CT medical image segmentation algorithm based on deep learning technology

Tongping Shen; Fangliang Huang; Xusong Zhang; Tongping Shen; Fangliang Huang; Xusong Zhang

doi:10.3934/mbe.2023485

Mathematical Biosciences and Engineering

2023, Volume 20, Issue 6: 10954-10976. doi: 10.3934/mbe.2023485

Previous Article Next Article

Research article Special Issues

CT medical image segmentation algorithm based on deep learning technology

1.
School of Information Engineering, Anhui University of Chinese Medicine, Hefei, 230012, China
2.
Graduate School, Angeles University Foundation, Angeles 2009, Philippines

Academic Editor: Yang Kuang

Received: 11 March 2023 Revised: 31 March 2023 Accepted: 16 April 2023 Published: 21 April 2023

For the problems of blurred edges, uneven background distribution, and many noise interferences in medical image segmentation, we proposed a medical image segmentation algorithm based on deep neural network technology, which adopts a similar U-Net backbone structure and includes two parts: encoding and decoding. Firstly, the images are passed through the encoder path with residual and convolutional structures for image feature information extraction. We added the attention mechanism module to the network jump connection to address the problems of redundant network channel dimensions and low spatial perception of complex lesions. Finally, the medical image segmentation results are obtained using the decoder path with residual and convolutional structures. To verify the validity of the model in this paper, we conducted the corresponding comparative experimental analysis, and the experimental results show that the DICE and IOU of the proposed model are 0.7826, 0.9683, 0.8904, 0.8069, and 0.9462, 0.9537 for DRIVE, ISIC2018 and COVID-19 CT datasets, respectively. The segmentation accuracy is effectively improved for medical images with complex shapes and adhesions between lesions and normal tissues.

Keywords:

Citation: Tongping Shen, Fangliang Huang, Xusong Zhang. CT medical image segmentation algorithm based on deep learning technology[J]. Mathematical Biosciences and Engineering, 2023, 20(6): 10954-10976. doi: 10.3934/mbe.2023485

Related Papers:

[1]	Shuyue Ma, Jiawei Sun, Huimin Yu . Global existence and stability of temporal periodic solution to non-isentropic compressible Euler equations with a source term. Communications in Analysis and Mechanics, 2023, 15(2): 245-266. doi: 10.3934/cam.2023013
[2]	Zhigang Wang . Serrin-type blowup Criterion for the degenerate compressible Navier-Stokes equations. Communications in Analysis and Mechanics, 2025, 17(1): 145-158. doi: 10.3934/cam.2025007
[3]	Hongxia Lin, Sabana, Qing Sun, Ruiqi You, Xiaochuan Guo . The stability and decay of 2D incompressible Boussinesq equation with partial vertical dissipation. Communications in Analysis and Mechanics, 2025, 17(1): 100-127. doi: 10.3934/cam.2025005
[4]	Yang Liu, Xiao Long, Li Zhang . Long-time dynamics for a coupled system modeling the oscillations of suspension bridges. Communications in Analysis and Mechanics, 2025, 17(1): 15-40. doi: 10.3934/cam.2025002
[5]	Fangyuan Dong . Multiple positive solutions for the logarithmic Schrödinger equation with a Coulomb potential. Communications in Analysis and Mechanics, 2024, 16(3): 487-508. doi: 10.3934/cam.2024023
[6]	Zhen Wang, Luhan Sun . The Allen-Cahn equation with a time Caputo-Hadamard derivative: Mathematical and Numerical Analysis. Communications in Analysis and Mechanics, 2023, 15(4): 611-637. doi: 10.3934/cam.2023031
[7]	Rui Sun, Weihua Deng . A generalized time fractional Schrödinger equation with signed potential. Communications in Analysis and Mechanics, 2024, 16(2): 262-277. doi: 10.3934/cam.2024012
[8]	Chunyou Sun, Junyan Tan . Attractors for a Navier–Stokes–Allen–Cahn system with unmatched densities. Communications in Analysis and Mechanics, 2025, 17(1): 237-262. doi: 10.3934/cam.2025010
[9]	Yue Pang, Xiaotong Qiu, Runzhang Xu, Yanbing Yang . The Cauchy problem for general nonlinear wave equations with doubly dispersive. Communications in Analysis and Mechanics, 2024, 16(2): 416-430. doi: 10.3934/cam.2024019
[10]	Yang Liu . Global attractors for a nonlinear plate equation modeling the oscillations of suspension bridges. Communications in Analysis and Mechanics, 2023, 15(3): 436-456. doi: 10.3934/cam.2023021

Abstract

1. Introduction

In this paper, we study the 2D steady compressible Prandtl equations in $\left\{ x > 0, y > 0 \right\}$ :

$\begin{align} \left\{\begin{aligned} &u\partial_{x}u+v\partial_{y}u-\frac{1}{\rho}\partial^{2}_{y}u = -\frac{\partial_{x}P(\rho)}{\rho},\\ &\partial_{x}(\rho u)+\partial_{y}(\rho v) = 0,\\ & u|_{x = 0} = u_{0}(y), \quad\lim\limits_{y\to \infty}u = U(x),\\ &u|_{y = 0} = v|_{y = 0} = 0, \end{aligned}\right. \end{align}$

(1.1)

where $(u, v)$ is velocity field, $\rho(x)$ and $U(x)$ are the traces at the boundary $\left\{y = 0\right\}$ of the density and the tangential velocity of the outer Euler flow. The states $\rho, U$ satisfy the Bernoulli law

$\begin{align} U\partial_{x}U+\frac{\partial_{x}P(\rho)}{\rho} = 0. \end{align}$

(1.2)

The pressure $P(\rho)$ is a strictly increasing function of $\rho$ with $0 < \rho_{0}\leq\rho \leq \rho_{1}$ for some positive constants $\rho_{0}$ and $\rho_{1}$ .

In this paper, we assume that the pressure satisfies the favorable pressure gradient $\partial_{x}P\leq0$ , which implies that

$\partial_{x}\rho\leq0.$

The boundary layer is a very important branch in fluid mechanics. Ludwig Prandtl ^[14] first proposed the related theory of the boundary layer in 1904. Since then, many scholars have devoted themselves to studying the mathematical theory of the boundary layer ^{[1,7,8,9,11,12,17,18,19,21,22,23,24,26,27]}. For more complex fluids, such as compressible fluids, one can refer to ^[19,20,28] and the references therein for more details. Here, for our purposes, we only list some relevant works.

There are three very natural problems about the steady boundary layer: (ⅰ) Boundary layer separation, (ⅱ) whether Oleinik's solutions are smooth up to the boundary for any $x > 0$ and (ⅲ) vanishing viscosity limit of the steady Navier-Stokes system. Next, we will introduce the relevant research progress in these three aspects. The separation of the boundary layer is one of the very important research contents in the boundary layer theory. ^[17]. The earliest mathematical theory in this regard was proposed by Caffarelli and E in an unpublished paper ^[25]. Their results show that the existence time $x^{*}$ of the solutions to the steady Prandtl equations in the sense of Oleinik is finite under the adverse pressure gradient. Moreover, the family $u_{\mu}(x, y) = \mu^{-\frac{1}{2}}u(x^{*}-\mu x, \mu^{\frac{1}{4}}y)$ is compact in $C^{0}(\mathbb{R}_{+}^{2})$ . Later, Dalibard and Masmoudi ^[4] proved the solution behaves near the separation as $\partial_{y} u(x, 0)\sim(x^{*}-x)^{\frac{1}{2}}$ for $x < x^{*}$ . Shen, Wang and Zhang ^[18] found that the solution near the separation point behaves like $\partial_{y} u(x, y)\sim(x^{*}-x)^{\frac{1}{4}}$ for $x < x^{*}$ . The above work further illustrates that the boundary layer separation is a very complex phenomenon. Recently, there were also some results about the steady compressible boundary layer separation ^[28]. The authors found that if the heat transfer in the boundary layer disappeared, then the singularity would be the same as that in the incompressible case. There is still relatively little mathematical theory on the separation of unsteady boundary layers. This is because back-flow and separation no longer occur simultaneously. When the boundary layer back-flow occurs, the characteristics of the boundary layer will continue to maintain for a period of time. Therefore, it is very important to study the back-flow point for further research on separation. Recently, Wang and Zhu ^[21] studied the back-flow problem of the two-dimensional unsteady boundary layer, which is a important work. It is very interesting to further establish the mathematical theory of the unsteady boundary layer separation.

Due to degenerate near the boundary, the high regularity of the solution of the boundary layer equation is a very difficult and meaningful work. In a local time $0 < x < x^{*}\ll 1$ , Guo and Iyer ^[6] studied the high regularity of of the Prandtl equations. Oleinik and Samokhin ^[13] studied the existence of solutions of steady Prandtl equations and Wang and Zhang ^[23] proved that Oleinik's solutions are smooth up to the boundary $y = 0$ for any $x > 0$ . The goal of this paper is to prove the global $C^{\infty}$ regularity of the two-dimensional steady compressible Prandtl equations. Recently, Wang and Zhang ^[24] found the explicit decay for general initial data with exponential decay by using the maximum principle.

In addition, in order to better understand the relevant background knowledge, we will introduce some other related work. As the viscosity goes to zero, the solutions of the three-dimensional evolutionary Navier-Stokes equations to the solutions of the Euler equations are an interesting problem. Beirão da Veiga and Crispo ^[2] proved that in the presence of flat boundaries convergence holds uniformly in time with respect to the initial data's norm. For the non-stationary Navier-Stokes equations in the 2D power cusp domain, the formal asymptotic expansion of the solution near the singular point is constructed and the constructed asymptotic decomposition is justified in ^[15,16] by Pileckas and Raciene.

Before introducing the main theorem, we introduce some preliminary knowledge. To use the von Mises transformation, we set

$\begin{align*} \tilde{u}(x,y) = \rho(x) u(x,y),\quad \tilde{v}(x,y) = \rho(x) v(x,y),\quad \tilde{u}_{0}(y) = \rho(0)u_{0}(y), \end{align*}$

then we find that $(\tilde{u}, \tilde{v})$ satisfies:

$\begin{align} \left\{\begin{aligned} &\tilde{u}\partial_{x}\tilde{u}+\tilde{v}\partial_{y}\tilde{u}-\partial^{2}_{y}\tilde{u}-\frac{\partial_{x}\rho}{\rho}\tilde{u}^{2} = -\rho \partial_{x}P(\rho),\\ &\partial_{x}\tilde{u}+\partial_{y}\tilde{v} = 0,\\ &\tilde{u}|_{x = 0} = \tilde{u}_{0}(y),\quad\lim\limits_{y\to \infty}\tilde{u} = \rho(x)U(x),\\ &\tilde{u}|_{y = 0} = \tilde{v}|_{y = 0} = 0. \end{aligned}\right. \end{align}$

(1.3)

By the von Mises transformation

$\begin{align} x = x, \quad \psi(x,y) = \int_{0}^{y}\tilde{u}(x,z)dz,\quad w = \tilde{u}^{2}, \end{align}$

(1.4)

$\begin{align} \begin{aligned} \partial_{x}\tilde{u} = \frac{\partial_{x}\omega}{2\sqrt{\omega}}+\frac{\partial_{\psi}\omega\partial_{x}\psi}{2\sqrt{\omega}}, \quad \partial_{y}\tilde{u} = \frac{\partial_{\psi}\omega}{2}, \quad \partial_{y}^{2}\tilde{u} = \frac{\sqrt{\omega}\partial_{\psi}^{2}\omega}{2}, \end{aligned} \end{align}$

(1.5)

and (1.3)–(1.5), we know that $w(x, \psi)$ satisfies:

$\begin{align} \partial_{x}w-\sqrt{w}\partial_{\psi}^{2}w-2\frac{\partial_{x}\rho}{\rho}w = -2\rho\partial_{x}P(\rho), \end{align}$

(1.6)

with

$\begin{align} w(x,0) = 0,\quad w(0,\psi) = w_{0}(\psi),\quad \lim\limits_{\psi\to+\infty}w = (\rho(x)U(x))^{2}. \end{align}$

(1.7)

In addition, we have

$\begin{align} 2\partial_{y}\tilde{u} = \partial_{\psi}w,\qquad2\partial_{y}^{2}\tilde{u} = \sqrt{w}\partial_{\psi}^{2}w. \end{align}$

(1.8)

In ^[5], Gong, Guo and Wang studied the existence of the solutions of system (1.1) by using the von Mises transformation and the maximal principle proposed by Oleinik and Samokhin in ^[13]. Specifically, they proved that:

Theorem 1.1. If the initial data $u_{0}$ satisfies the following conditions:

$\begin{align} \begin{aligned} &u\in C^{2,\alpha}_{b}([0,+\infty))(\alpha > 0),\quad u(0) = 0,\quad\partial_{y}u(0) > 0,\quad\partial_{y}u(y)\geq0\quad\mathit{\text{for}}\quad y\in[0,+\infty), \\ &\lim\limits_{y\to+\infty}u(y) = U(0) > 0,\quad\rho^{-1}(0)\partial_{y}^{2}u(y)-\rho^{-1}(0)\partial_{x}P(0) = O(y^{2}) \end{aligned} \end{align}$

(1.9)

and $\rho\in C^{2}([0, X_{0}])$ , then there exists $0 < X\leq X_{0}$ such that system (1.1) admits a solution $u\in C^{1}([0, X)\times \mathbb{R}_{+})$ . The solution has the following properties:

$(i)$ $u$ is continuous and bounded in $[0, X]\times \mathbb{R}_{+}$ ; $\partial_{y}u, \partial_{y}^{2}u$ are continuous and bounded in $[0, X)\times \mathbb{R}_{+}$ ; $v, \partial _{y}v, \partial_{x}u$ are locally bounded in $[0, X)\times \mathbb{R}_{+}$ .

$(ii)$ $u(x, y) > 0$ in $[0, X)\times \mathbb{R}_{+}$ and for any $\bar{x} < X$ , there exists $y_{0}, m > 0$ such that for all $(x, y)\in[0, \bar{x}]\times[0, y_{0}]$ ,

$\partial_{y}u(x,y )\geq m > 0.$

$(iii)$ if $\partial_{x}P\leq0(\partial_{x}\rho\leq0)$ , then

$X = +\infty.$

Remarks 1.2. $u\in C^{2, \alpha}_{b}([0, +\infty))(\alpha > 0)$ means that $u$ is Hölder continuity and bounded.

This theorem shows that under the favorable pressure gradient, the solution is global-in- $x$ . However, if the pressure is an adverse pressure gradient, then boundary layer separation will occur. Xin and Zhang ^[26] studied the global existence of weak solutions of unsteady Prandtl equations under the favorable pressure gradient. For the unsteady compressible Prandtl equation, similar results are obtained in ^[3]. Recently, Xin, Zhang and Zhao ^[27] proposed a direct proof of the existence of global weak solutions of the Prandtl equation. The key content of this paper is that they have studied the uniqueness and regularity of weak solutions. This method can be applied to the compressible Prandtl equation.

Our main results are as follows:

Theorem 1.3. If $u$ is a solution for equation (1.1) in Theorem 1.1, assume $u_{0}$ satisfies the condition (1.9) and the known function $\rho$ and $\partial_{x}P$ are smooth. Then, there exists a constant $C > 0$ depending only on $\varepsilon, X, u_{0}, P(\rho), k, m$ such that for any $(x, y)\in[\varepsilon, X]\times[0, +\infty),$

$\begin{align*} |\partial_{x}^{k}\partial_{y}^{m}u(x,y)|\leq C, \end{align*}$

where $X, \varepsilon$ are positive constants with $\varepsilon < X$ and $m, k$ are any positive integers.

Remarks 1.4. Our methods may be used to other related models. There are similar results for the magnetohydrodynamics boundary layer and the thermal boundary layer. This work will be more difficult due to the influence of temperature and the magnetic field.

Due to the degeneracy near the boundary $\psi = 0$ , the proof of the main result is divided into two parts, Theorem 1.5 and Theorem 1.6. This is similar to the result of the incompressible boundary layer, despite the fluid being compressible and the degeneracy near the boundary. Different from the incompressible case ^[23], we have no divergence-free conditions, which will bring new terms. It is one of the difficulties in this paper to deal with these terms. Now, we will briefly introduce our proof framework. First, we prove the following theorem in the domain $[\varepsilon, X]\times[0, Y_{1}]$ for a small $Y_{1}$ . The key ingredients of proof is that we employ interior priori estimates and the maximum principle developed by Krylov ^[10].

Theorem 1.5. If $u$ is a solution for equation (1.1) in Theorem 1.1, assume $u_{0}$ satisfies the condition (1.9) and the known function $\rho$ and $\partial_{x}P$ are smooth. Then, there exists a small constant $Y_{1} > 0$ and a large constant $C > 0$ depending only on $\varepsilon, X, Y_{1}, u_{0}, P(\rho), k, m$ such that for any $(x, y)\in[\varepsilon, X]\times[0, Y_{1}],$

$\begin{align*} |\partial_{x}^{k}\partial_{y}^{m}u(x,y)|\leq C, \end{align*}$

where $X, \varepsilon$ are positive constants with $\varepsilon < X$ and $m, k$ are any positive integers.

Next, we prove the following theorem in the domain $[\varepsilon, X]\times[Y_{2}, +\infty)$ for a small positive constant $Y_{2}$ . The key of proof is that we prove (1.6) is a uniform parabolic equation in the domain $[\varepsilon, X]\times[Y_{2}, +\infty)$ in Section 4. Once we have (1.6) is a uniform parabolic equation, the global $C^{\infty}$ regularity of the solution is a direct result of interior Schauder estimates and classical parabolic regularity theory. The proof can be given similarly to the steady incompressible boundary layer. For the sake of simplicity of the paper, more details can be found in ^[23] and we omit it here.

Theorem 1.6. If $u$ is a solution for equation (1.1) in Theorem 1.1, assume $u_{0}$ satisfies the condition (1.9) and the known function $\rho$ and $\partial_{x}P$ are smooth. Then, there exists a constant $Y_{0} > 0$ such that for any constant $Y_{2}\in(0, Y_{0})$ , there exists a constant $C > 0$ depending only on $\varepsilon, X, Y_{2}, u_{0}, P(\rho), k, m$ such that for any $(x, y)\in[\varepsilon, X]\times[Y_{2}, +\infty),$

$\begin{align*} |\partial_{x}^{k}\partial_{y}^{m}u(x,y)|\leq C, \end{align*}$

where $X, \varepsilon$ are positive constants with $\varepsilon < X$ and $m, k$ are any positive integers.

Therefore, Theorem 1.3 can be directly proven by combining Theorem 1.5 with Theorem 1.6.

The organization of this paper is as follows. In Section 2, we study lower order and higher order regularity estimates. In Section 3, we prove Theorem 1.5 in the domain near $y = 0$ by transforming back to the original coordinates $(x, y)$ . In Section 4, we prove (1.6) is a uniform parabolic equation by using the maximum principle and we also prove the Theorem 1.3.

2. Lower order and higher order regularity estimates

2.1. Lower order regularity estimates

In this subsection, we study the lower order regularity estimates using the standard interior a priori estimates developed by Krylov ^[10].

Lemma 2.1. If $u$ is a solution for equation (1.1) in Theorem 1.1, assume $u_{0}$ satisfies the condition (1.9)and the known function $\rho$ and $\partial_{x}P$ are smooth. Assume $0 < \varepsilon < X$ , then there exists some positive constants $\delta_{1} > 0$ and $C$ independent of $\psi$ such that for any $(x, \psi)\in[\varepsilon, X]\times[0, \delta_{1}],$

$\begin{align*} |\partial_{x}w(x,\psi)|\leq C\psi. \end{align*}$

Proof. Due to Lemma 2.1 in ^[5] (or Theorem 2.1.14 in ^[13]), there exists $\delta_{1} > 0$ for any $(x, \psi)\in[0, X]\times[0, \delta_{1}]$ , such that for some $\alpha\in(0, \frac{1}{2})$ and positive constants $m, M$ (we assume $\delta_{1} < 1$ ),

$\begin{align} |\partial_{x}w|\leq C\psi^{\frac{1}{2}+\alpha},\quad 0 < m < \partial_{\psi}w < M,\quad m\psi < w < M\psi. \end{align}$

(2.1)

By (1.6), we obtain

$\begin{align*} \partial_{x} \partial_{x} w-\sqrt{w} \partial_{\psi}^{2} \partial_{x} w = \frac{\left(\partial_{x} w\right)^{2}}{2 w}+2 \frac{\rho\partial_{x}P \partial_{x}w }{2 w}+\frac{\partial_{x}\rho}{\rho}\partial_{x}w+2\partial_{x}\left(\frac{\partial_{x}\rho}{\rho}\right)w-2 \partial_{x}[\rho\partial_{x}P]. \end{align*}$

Take a smooth cutoff function $0 \leq \phi(x) \leq 1$ in $[0, X]$ such that

$\begin{align*} \phi(x) = 1, x \in [\varepsilon,X],\quad \phi(x) = 0, x \in[0,\frac{\varepsilon}{2}], \end{align*}$

then

$\begin{align*} \begin{aligned} \partial_{x}\left[\partial_{x} w \phi(x)\right]&-\sqrt{w} \partial_{\psi}^{2}\left[\partial_{x} w \phi(x)\right]\\ & = \frac{\left(\partial_{x} w\right)^{2}}{2 w} \phi(x)+2 \frac{\rho\partial_{x}P \partial_{x}w }{2 w}\phi(x)+\frac{\partial_{x}\rho}{\rho}\partial_{x}w\phi(x)\\ &+2\partial_{x}\left(\frac{\partial_{x}\rho}{\rho}\right)w\phi(x) -2 \partial_{x}(\rho\partial_{x}P)\phi(x)+\partial_{x} w \partial_{x} \phi(x) : = \mathcal{W}. \end{aligned} \end{align*}$

Combining with (2.1), we know

$\begin{align} |\mathcal{W}| \leq C \psi^{2 \alpha}+C \psi^{\alpha-\frac{1}{2}}+C\psi^{\alpha+\frac{1}{2}}+C\psi+C \leq C \psi^{\alpha-\frac{1}{2}}. \end{align}$

(2.2)

We take $\varphi(\psi) = \mu_{1} \psi-\mu_{2} \psi^{1+\beta}$ with constants $\mu_{1}, \mu_{2}$ , then by (2.1) and (2.2), we get

$\begin{align*} \begin{aligned} \partial_{x}\left[\partial_{x} w \phi(x)-\varphi\right]-\sqrt{w} \partial_{\psi}^{2}\left[\partial_{x} w \phi(x)-\varphi\right] & \leq|\mathcal{W}|-\mu_{2} \sqrt{w} \beta(1+\beta) \psi^{\beta-1} \\ & \leq C \psi^{\alpha-\frac{1}{2}}-\mu_{2} \sqrt{m} \beta(1+\beta) \psi^{\beta-\frac{1}{2}}. \end{aligned} \end{align*}$

By taking $\mu_{2}$ sufficiently large and $\alpha = \beta$ , for $(x, \psi)\in(0, X] \times\left(0, \delta_{1}\right)$ , we have

$\begin{align*} \partial_{x}\left[\partial_{x} w \phi(x)-\varphi\right]-\sqrt{w} \partial_{\psi}^{2}\left[\partial_{x} w \phi(x)-\varphi\right] < 0. \end{align*}$

For any $\psi\in\left[0, \delta_{1}\right]$ , let $\mu_{1} \geq \mu_{2}$ , and we have

$\begin{align*} \left(\partial_{x} w \phi-\varphi\right)(0, \psi) \leq 0 , \end{align*}$

and take $\mu_{1}$ large enough depending on $M, \delta_{1}, \mu_{2}$ such that

$\begin{align*} \left(\partial_{x} w \phi-\varphi\right)\left(x, \delta_{1}\right)\leq M \delta_{1}^{\frac{1}{2}+\alpha}-\mu_{1} \delta_{1}+\mu_{2} \delta_{1}^{1+\beta} \leq 0 . \end{align*}$

Since $w(x, 0) = 0$ , we know that for any $x\in[0, X]$ ,

$\begin{align*} \left(\partial_{x} w \phi-\varphi\right)(x, 0) = 0. \end{align*}$

By the maximum principle, it holds in $[0, X] \times[0, \delta_{1}]$ that

$\begin{align*} \left(\partial_{x} w \phi-\varphi\right)(x, \psi) \leq 0. \end{align*}$

Let $\delta_{1}$ be chosen suitably small, for $(x, \psi)\in[\varepsilon, X] \times[0, \delta_{1}]$ , and we obtain

$\begin{align*} \partial_{x} w(x, \psi) \leq \mu_{1} \psi-\mu_{2} \psi^{1+\beta} \leq \frac{\mu_{1}}{2} \psi. \end{align*}$

Considering $-\partial_{x} w \phi-\varphi$ , the result $-\partial_{x} w \leq \frac{\mu_{1}}{2} \psi$ in $[\varepsilon, X] \times[0, \delta_{1}]$ can be proved similarly. This completes the proof of the lemma.

Lemma 2.2. If $u$ is a solution for equation (1.1) in Theorem 1.1, assume $u_{0}$ satisfies the condition (1.9) and the known function $\rho$ and $\partial_{x}P$ are smooth. Assume $0 < \varepsilon < X$ , then there exists some positive constants $\delta_{2} > 0$ and $C$ independent of $\psi$ such that for any $(x, \psi)\in[\varepsilon, X]\times[0, \delta_{2}],$

$\begin{align*} |\partial_{\psi} \partial_{x} w(x, \psi)| \leq C, \quad|\partial_{x}^{2} w(x, \psi)| \leq C \psi^{-\frac{1}{2}}, \quad|\partial_{\psi}^{2} \partial_{x} w(x, \psi)| \leq C \psi^{-1}. \end{align*}$

Proof. From Lemma 2.1, there exists $\delta_{1} > 0$ such that for any $(x, \psi)\in[\frac{\varepsilon}{2}, X] \times[0, \delta_{1}],$

$\begin{align*} |\partial_{x} w(x, \psi)| \leq C \psi. \end{align*}$

Let $\Psi_{0} = \min \{\frac{2}{3} \delta_{1}, \frac{\varepsilon}{2}\}$ , for any $(x_{0}, \psi_{0}) \in[\varepsilon, X] \times(0, \Psi_{0}]$ , and we denote

$\begin{align*} \Omega = \bigl\{(x,\psi)|x_{0}-\psi_{0}^{\frac{3}{2}} \leq x \leq x_{0},\frac{1}{2} \psi_{0} \leq \psi \leq \frac{3}{2} \psi_{0}\bigr\}. \end{align*}$

By the definition of $\Psi_{0}$ , we know $\Omega\subseteq[\frac{\varepsilon}{2}, X] \times[0, \delta_{1}]$ , then it holds in $\Omega$ that

$\begin{align} |\partial_{x} w| \leq C \psi. \end{align}$

(2.3)

The following transformation $f$ is defined:

$\begin{align*} \begin{aligned} \Omega \rightarrow \widetilde{\Omega}: = [-1,0]_{\tilde{x}} \times[-\frac{1}{2}, \frac{1}{2}]_{\tilde{\psi}}, \quad (x, \psi) \mapsto(\tilde{x}, \tilde{\psi}), \end{aligned} \end{align*}$

where $x-x_{0} = \psi_{0}^{\frac{3}{2}} \tilde{x}, \psi-\psi_{0} = \psi_{0} \tilde{\psi}$ .

Since $\partial_{\tilde{x}} = \psi_{0}^{\frac{3}{2}}\partial_{x}, \partial_{\tilde{\psi}} = \psi_{0} \partial_{\psi}$ , it holds in $\Omega$ that

$\begin{align*} \partial_{\tilde{x}}\left(\psi_{0}^{-1} w\right)-\psi_{0}^{-\frac{1}{2}} \sqrt{w} \partial_{\tilde{\psi}}^{2}\left(\psi_{0}^{-1} w\right)-2\frac{\partial_{\tilde{x}}\rho}{\rho}(\psi_{0}^{-1}w) = -2 \rho\partial_{\tilde{x}}P \psi_{0}^{-1} . \end{align*}$

Combining with (2.1), we get $0 < c \leq\psi_{0}^{-\frac{1}{2}} \sqrt{w} \leq C, |\psi_{0}^{-1} w| \leq C$ , and for any $\widetilde{z}_{1}, \widetilde{z}_{2} \in \widetilde{\Omega}$ ,

$\begin{align*} \begin{aligned} |\psi_{0}^{-\frac{1}{2}} \sqrt{w}(\tilde{z}_{1})-\psi_{0}^{-\frac{1}{2}} \sqrt{w}(\tilde{z}_{2})| = \psi_{0}^{-\frac{1}{2}} \frac{\left|w\left(\tilde{z}_{1}\right)-w\left(\tilde{z}_{2}\right)\right|}{\sqrt{w}\left(\tilde{z}_{1}\right)+\sqrt{w}\left(\tilde{z}_{2}\right)} \leq C \frac{\psi_{0}\left|\tilde{z}_{1}-\tilde{z}_{2}\right|}{\psi_{0}} = C\left|\tilde{z}_{1}-\tilde{z}_{2}\right|. \end{aligned} \end{align*}$

This means that for any $\alpha \in(0, 1)$ , we have

$\begin{align*} |\psi_{0}^{-\frac{1}{2}}\sqrt{w}|_{\mathcal{C}^{\alpha}(\widetilde{\Omega})} \leq C. \end{align*}$

Since $P$ and $\rho$ are smooth, we have

$\begin{align*} |\rho^{-1}\partial_{\tilde{x}}\rho|_{\mathcal{C}^{0,1}([-1,0]_{\tilde{x}})}+|\rho\partial_{\tilde{x}}P\psi_{0}^{-1}|_{\mathcal{C}^{0,1}([-1,0]_{\tilde{x}})} \leq C . \end{align*}$

By standard interior priori estimates (see Theorem 8.11.1 in ^[10] or Proposition 2.3 in ^[23]), we have

$\begin{align} | w \psi_{0}^{-1}|_{\mathcal{C}^{\alpha}([-\frac{1}{2}, 0]_{\tilde{x}} \times[-\frac{1}{4}, \frac{1}{4}]_{\tilde{\psi}})}+|\partial_{\tilde{\psi}}^{2} w \psi_{0}^{-1}|_{\mathcal{C}^{\alpha}([-\frac{1}{2}, 0]_{\tilde{x}} \times[-\frac{1}{4}, \frac{1}{4}]_{\tilde{\psi}})} \leq C. \end{align}$

(2.4)

Let $f: = \partial_{x} w \psi_{0}^{-1}$ , which satisfies

$\begin{align*} \partial_{\tilde{x}} f-\frac{\sqrt{w}}{\psi_{0}^{\frac{1}{2}}} \partial_{\tilde{\psi}}^{2} f-\frac{\partial_{\tilde{\psi}}^{2} w}{2 \sqrt{w} \psi_{0}^{\frac{1}{2}}} f-2\frac{\partial_{\tilde{x}}\rho}{\rho}f = -2 \partial_{x}[\rho\partial_{\tilde{x}}P] \psi_{0}^{-1}+2\partial_{x}\left(\frac{\partial_{\tilde{x}}\rho}{\rho}\right)(\psi_{0}^{-1}w) . \end{align*}$

By (2.3), we have $|f| \leq C$ in $\widetilde{\Omega}$ . Due to

$\begin{align*} \big|\psi_{0}^{\frac{1}{2}}w^{-\frac{1}{2}}(\tilde{z}_{1})-\psi_{0}^{\frac{1}{2}}w^{-\frac{1}{2}}(\tilde{z}_{2})\big| = \psi_{0}^{\frac{1}{2}} \frac{\left|\frac{w(\tilde{z}_{1})-w(\tilde{z}_{2})}{w(\tilde{z}_{1}) w(\tilde{z}_{2})}\right|}{w^{-\frac{1}{2}}(\tilde{z}_{1})+w^{-\frac{1}{2}}(\tilde{z}_{2})} \leq C\left|\tilde{z}_{1}-\tilde{z}_{2}\right|, \end{align*}$

we have

$\begin{align} \big|\psi_{0}^{\frac{1}{2}}w^{-\frac{1}{2}}\big|_{\mathcal{C}^{\alpha}(\widetilde{\Omega})} \leq C. \end{align}$

(2.5)

Since

$\frac{\partial_{\tilde{\psi}}^{2} w}{2 \sqrt{w} \psi_{0}^{\frac{1}{2}}} = \partial_{\tilde{\psi}}^{2} w \psi_{0}^{-1} \frac{\psi_{0}^{\frac{1}{2}}}{2 \sqrt{w}},$

which along with (2.4) and (2.5) gives

$\begin{align*} \bigg|\frac{\partial_{\tilde{\psi}}^{2} w}{2 \sqrt{w} \psi_{0}^{\frac{1}{2}}}\bigg|_{\mathcal{C}^{\alpha}([-\frac{1}{2}, 0]_{\tilde{x}} \times[-\frac{1}{4}, \frac{1}{4}]_{\tilde{\psi}})} \leq C . \end{align*}$

As before, by (2.4) and the density $\rho$ and $P$ are smooth, via the standard interior a priori estimates, it yield that

$\begin{align*} |\partial_{\tilde{x}} f|_{L^{\infty}([-\frac{1}{4}, 0]_{\tilde{x}} \times[-\frac{1}{8}, \frac{1}{8}]_{\tilde{\psi}})}+|\partial_{\tilde{\psi}} f|_{L^{\infty}([-\frac{1}{4}, 0]_{\tilde{x}} \times[-\frac{1}{8}, \frac{1}{8}]_{\tilde{\psi}})}+|\partial_{\tilde{\psi}}^{2} f|_{L^{\infty}([-\frac{1}{4}, 0]_{\tilde{x}} \times[-\frac{1}{8}, \frac{1}{8}]_{\tilde{\psi}})} \leq C . \end{align*}$

Therefore, we obtain

$\begin{align*} |\partial_{x}^{2} w\left(x_{0}, \psi_{0}\right)| \leq C \psi_{0}^{-\frac{1}{2}}, \quad|\partial_{\psi} \partial_{x} w\left(x_{0}, \psi_{0}\right)| \leq C, \quad|\partial_{\psi}^{2} \partial_{x} w\left(x_{0}, \psi_{0}\right)| \leq C \psi_{0}^{-1}. \end{align*}$

This completes the proof of the lemma.

2.2. Higher order regularity estimates

In this subsection, we study the higher order regularity estimates using the maximum principle. The two main results of this subsection are Lemma 2.3 and Lemma 2.7.

Lemma 2.3. If $u$ is a solution for equation (1.1) in Theorem 1.1, assume $u_{0}$ satisfies the condition (1.9) and the known function $\rho$ and $\partial_{x}P$ are smooth. Assume $0 < \varepsilon < X$ and $k \geq 2$ , then there exists some positive constants $\delta > 0$ and $C$ independent of $\psi$ such that for any $(x, \psi)\in[\varepsilon, X]\times[0, \delta],$

$\begin{align*} |\partial_{x}^{k} w| \leq C \psi, \quad|\partial_{\psi} \partial_{x}^{k} w| \leq C, \quad|\partial_{\psi}^{2} \partial_{x}^{k} w| \leq C \psi^{-1} . \end{align*}$

Proof. By Lemma 2.1 and Lemma 2.2, we may inductively assume that for $0 \leq j \leq k-1$ , there holds that in $\left[\frac{\varepsilon}{2}, X\right] \times\left[0, \delta_{3}\right]$ (assume $\delta_{3}\ll1$ ),

$\begin{align} |\partial_{\psi} \partial_{x}^{j} w| \leq C,\quad|\partial_{\psi}^{2} \partial_{x}^{j} w| \leq C \psi^{-1},\quad|\partial_{x}^{j} w| \leq C \psi,\quad|\partial_{x}^{j} \sqrt{w}| \leq C \psi^{\frac{1}{2}},\quad|\partial_{x}^{k} w| \leq C \psi^{-\frac{1}{2}}. \end{align}$

(2.6)

We will prove that there exists $\delta_{4} < \delta_{3}$ so that in $[\varepsilon, X] \times[0, \delta_{4}]$ ,

$\begin{align} |\partial_{\psi} \partial_{x}^{k} w| \leq C,\quad|\partial_{\psi}^{2} \partial_{x}^{k} w| \leq C \psi^{-1},\quad|\partial_{x}^{k} w| \leq C \psi,\quad|\partial_{x}^{k} \sqrt{w}| \leq C \psi^{\frac{1}{2}},\quad|\partial_{x}^{k+1} w| \leq C \psi^{-\frac{1}{2}}. \end{align}$

(2.7)

The above results are deduced from the following Lemma 2.4, Lemma 2.5 and Lemma 2.6.

Lemma 2.4. If $u$ is a solution for equation (1.1) in Theorem 1.1, assume $u_{0}$ satisfies the condition (1.9) and the known function $\rho$ and $\partial_{x}P$ are smooth. Assume that (2.6) holds, then there is a positive constant $M_{1}$ for any $(x, \psi)\in\left[\frac{7 \varepsilon}{8}, X\right] \times\left[0, \delta_{3}\right]$ and $0 < \beta \ll 1$ ,

$\begin{align*} |\partial_{x}^{k} w| < M_{1} \psi^{1-\beta}, \quad|\partial_{x}^{k} \sqrt{w}| \leq M_{1} \psi^{\frac{1}{2}-\beta}. \end{align*}$

Proof. Take a smooth cutoff function $0 \leq \phi(x) \leq 1$ in $[0, X]$ such that

$\begin{align*} \phi(x) = 1, x \in [\frac{7\varepsilon}{8},X],\quad \phi(x) = 0, x \in[0,\frac{5\varepsilon}{8}]. \end{align*}$

As in ^[23], fix any $h < \frac{\varepsilon}{8}$ . Set

$\Omega = \left\{(x,\psi)|0 < x\leq X, 0 < \psi < \delta_{3}\right\},$

and let

(ⅰ) $f = \frac{\partial_{x}^{k-1} w(x-h, \psi)-\partial_{x}^{k-1} w(x, \psi)}{-h} \phi+M \psi \ln \psi,$ $(x, \psi)\in[\frac{5 \varepsilon}{8}, X]\times [0, +\infty)$ ,

(ⅱ) $f = M \psi \ln \psi,$ $(x, \psi)\in[0, \frac{5 \varepsilon}{8})\times[\psi, +\infty),$

so we get $f(x, 0) = 0$ , $f(0, \psi) \leq 0$ . We know

$f\left(x, \delta_{3}\right) \leq C\left(\delta_{3}\right)^{-\frac{1}{2}}+M \delta_{3} \ln \delta_{3} \leq 0,$

where $M$ is large enough. Then, by choosing the appropriate $M$ , we know that the positive maximum of $f$ cannot be achieved in the interior. Finally, the lemma can be proven by the arbitrariness of $h$ .

Assume that there exists a point

$p_{0} = (x_{0}, \psi_{0}) \in \Omega,$

such that

$f\left(p_{0}\right) = \max \limits_{\bar{\Omega}} f > 0.$

It is easy to know that

$x_{0} > \frac{5 \varepsilon}{8},\qquad \partial_{x}^{k-1} w(x_{0}-h, \psi_{0}) < \partial_{x}^{k-1} w(x_{0}, \psi_{0}).$

By (2.1), denote $\xi = \sqrt{m}$ , we have

$\begin{align} -\sqrt{w} \partial_{\psi}^{2}(M \psi \ln \psi) = -M \sqrt{w} \psi^{-1} \leq-\xi M \psi^{-\frac{1}{2}}. \end{align}$

(2.8)

By (1.6), a direct calculation gives

$\begin{align*} \begin{aligned} \partial_{x} \partial_{x}^{k-1} w-&\sqrt{w} \partial_{\psi}^{2} \partial_{x}^{k-1} w\\ & = -2 \partial_{x}^{k-1}(\rho\partial_{x}P)+\sum\limits_{m = 1}^{k-2} C_{k-1}^{m}\left(\partial_{x}^{k-1-m} \sqrt{w}\right) \partial_{\psi}^{2} \partial_{x}^{m} w+\left(\partial_{x}^{k-1} \sqrt{w}\right) \partial_{\psi}^{2} w \\ &\quad+2\sum\limits_{m = 0}^{k-1} C_{k-1}^{m}\partial_{x}^{k-1-m}\left(\frac{\partial_{x}\rho}{\rho}\right)\partial_{x}^{m}w\\ & = -2 \partial_{x}^{k-1}(\rho\partial_{x}P)+\sum\limits_{m = 1}^{k-2} C_{k-1}^{m}\left(\partial_{x}^{k-1-m} \sqrt{w}\right) \partial_{\psi}^{2} \partial_{x}^{m} w+\frac{\partial_{x}^{k-1} w}{2 \sqrt{w}} \frac{\partial_{x} w}{\sqrt{w}} \\ &\quad+\left(\frac{\partial_{x}^{k-1} w}{2 \sqrt{w}}\right) \frac{2\rho\partial_{x}P}{\sqrt{w}}-\left(\frac{\partial_{x}^{k-1} w}{2 \sqrt{w}}\right)\frac{2\frac{\partial_{x}\rho}{\rho}w}{\sqrt{w}}+2\sum\limits_{m = 0}^{k-1} C_{k-1}^{m}\partial_{x}^{k-1-m}\left(\frac{\partial_{x}\rho}{\rho}\right)\partial_{x}^{m}w\\ &\quad+ \sum\limits_{m = 0}^{k-2} C_{k-2}^{m} \partial_{\psi}^{2} w \partial_{x}^{m+1} w \partial_{x}^{k-2-m} \frac{1}{2 \sqrt{w}}\\ &: = \sum\limits_{i = 1}^{4}I_{i} \end{aligned} \end{align*}$

and

$\begin{align*} \begin{aligned} I_{1} & = -2 \partial_{x}^{k-1}(\rho\partial_{x}P)+\sum\limits_{m = 1}^{k-2} C_{k-1}^{m}\left(\partial_{x}^{k-1-m} \sqrt{w}\right) \partial_{\psi}^{2} \partial_{x}^{m} w+\frac{\partial_{x}^{k-1} w}{2 \sqrt{w}} \frac{\partial_{x} w}{\sqrt{w}}, \\ I_{2} & = \frac{\rho\partial_{x}P}{w}\partial_{x}^{k-1} w, \\ I_{3} & = -\frac{\partial_{x}\rho}{\rho}\partial_{x}^{k-1}w+2\sum\limits_{m = 0}^{k-1} C_{k-1}^{m}\partial_{x}^{k-1-m}\left(\frac{\partial_{x}\rho}{\rho}\right)\partial_{x}^{m}w,\\ I_{4}& = \sum\limits_{m = 0}^{k-2} C_{k-2}^{m} \partial_{\psi}^{2} w \partial_{x}^{m+1} w \partial_{x}^{k-2-m} \frac{1}{2 \sqrt{w}}. \end{aligned} \end{align*}$

For $x \geq \frac{5 \varepsilon}{8}$ , we consider the following equality

$\begin{align} \partial_{x} f_{1}-\sqrt{w\left(p_{1}\right)} \partial_{\psi}^{2} f_{1} = \frac{\sqrt{w(p_{1})}-\sqrt{w(p)}}{-h} \partial_{\psi}^{2} \partial_{x}^{k-1} w(p)+\sum\limits_{i = 1}^{4} \frac{1}{-h}\left(I_{i}\left(p_{1}\right)-I_{i}(p)\right), \end{align}$

(2.9)

where

$f_{1} = \frac{1}{-h}(\partial_{x}^{k-1} w\left(p_{1}\right)-\partial_{x}^{k-1} w(p)),$

with $p_{1} = (x-h, \psi), p = (x, \psi)$ .

For any $x \geq \frac{5 \varepsilon}{8}$ , by (2.6), it is easy to conclude that

$\begin{align} \begin{aligned} &\left|\frac{1}{-h}(\sqrt{w}(p_{1})-\sqrt{w}(p)) \partial_{\psi}^{2} \partial_{x}^{k-1} w(p)\right| \leq C \psi^{-\frac{1}{2}}, \\ &\left|\frac{1}{-h}(I_{1}(p_{1})-I_{1}(p))\right| \leq C \psi^{-\frac{1}{2}}, \\ &\left|\sum\limits_{i = 3}^{4}\frac{1}{-h}(I_{i}(p_{1})-I_{i }(p))\right| \leq C \psi^{-\frac{1}{2}}, \end{aligned} \end{align}$

(2.10)

where $C$ is dependent on the parameter $h$ .

Since

$\begin{align*} \frac{1}{-h}(I_{2}(p_{1})-I_{2}(p)) = f_{1} \cdot\big[\frac{\rho\partial_{x}P}{w} (p_{1})\big]+\partial_{x}^{k-1} w(p) \frac{1}{-h}\big[\frac{\rho\partial_{x}P}{w} (p_{1})-\frac{\rho\partial_{x}P}{w} (p)\big], \end{align*}$

combining with (2.6), $f_{1}(p_{0}) > 0$ and $\partial_{x}P\leq 0$ , it holds at $p = p_{0}$ that

$\begin{align} \frac{1}{-h}(I_{2}\left(p_{1}\right)-I_{2}(p_{0})) \leq C . \end{align}$

(2.11)

Summing up (2.10) and (2.11), we conclude that at $p = p_{0}$ ,

$\begin{align*} \partial_{x} f_{1}-\sqrt{w} \partial_{\psi}^{2} f_{1} \leq C_{0} \psi^{-\frac{1}{2}} . \end{align*}$

This along with (2.8) shows that for $x \geq \frac{5 \varepsilon}{8}$ , it holds at $p = p_{0}$ that

$\begin{align} \partial_{x} f-\sqrt{w} \partial_{\psi}^{2} f \leq C \psi^{-\frac{1}{2}}-\xi M \psi^{-\frac{1}{2}} . \end{align}$

(2.12)

By taking $M$ large enough, we have $\partial_{x} f\left(p_{0}\right)-\sqrt{w} \partial_{\psi}^{2} f\left(p_{0}\right) < 0$ . By the definition of $p_{0}$ , we obtain

$\begin{align*} \partial_{x} f\left(p_{0}\right)-\sqrt{w} \partial_{\psi}^{2} f\left(p_{0}\right) \geq 0, \end{align*}$

which leads to a contradiction. Therefore, for $M$ chosen as above and independent of $h$ , we have

$\max\limits _{\bar{\Omega}} f \leq 0.$

We can similarly prove that $\min _{\bar{\Omega}} f \geq 0$ by replacing $M \psi \ln \psi$ in $f$ with $-M \psi \ln \psi$ . By the arbitrariness of $h$ , for any $(x, \psi)\in(\frac{7\varepsilon}{8}, X] \times(0, \delta_{3}]$ we have

$\begin{align*} |\partial_{x}^{k} w| \leq-M \psi \ln \psi. \end{align*}$

Due to

$\begin{align} 2 \sqrt{w} \partial_{x}^{k} \sqrt{w}+\sum\limits_{m = 1}^{k-1} C_{k}^{m}(\partial_{x}^{m} \sqrt{w} \partial_{x}^{k-m} \sqrt{w}) = \partial_{x}^{k}(\sqrt{w} \sqrt{w}) = \partial_{x}^{k} w, \end{align}$

(2.13)

which along with (2.6) shows that in $(\frac{7}{8} \varepsilon, X] \times(0, \delta_{3}]$ ,

$\begin{align*} |\sqrt{w} \partial_{x}^{k} \sqrt{w}| \leq-C \psi \ln \psi . \end{align*}$

Lemma 2.5. If $u$ is a solution for equation (1.1) in Theorem 1.1, assume $u_{0}$ satisfies the condition (1.9) and the known function $\rho$ and $\partial_{x}P$ are smooth. Assume that (2.6) holds, then for any $(x, \psi)\in[\frac{15}{16} \varepsilon, X] \times[0, \delta_{3}]$ ,

$\begin{align*} |\partial_{x}^{k} w| \leq C \psi, \quad|\partial_{x}^{k} \sqrt{w}| \leq C \psi^{\frac{1}{2}} . \end{align*}$

Proof. Take a smooth cutoff function $\phi(x)$ so that

$\begin{align*} \phi(x) = 1, x \in [\frac{15\varepsilon}{16},X],\quad \phi(x) = 0, x \in[0,\frac{7\varepsilon}{8}]. \end{align*}$

Set

$f = \partial_{x}^{k} w \phi-\mu_{1} \psi+\mu_{2} \psi^{\frac{3}{2}-\beta}$

with constants $\mu_{1}, \mu_{2}$ . Let $\beta$ be small enough in Lemma 2.4. Then it holds in $[\frac{7 \varepsilon}{8}, X] \times[0, \delta_{3}]$ that

$\begin{align} |\partial_{x}^{k} w| \leq C \psi^{1-\beta}, \quad|\partial_{x}^{k} \sqrt{w}| \leq C \psi^{\frac{1}{2}-\beta}. \end{align}$

(2.14)

We denote

$\Omega = \left\{(x,\psi)|0 < x\leq X ,0 < \psi < \delta_{3}\right\}.$

As in ^[23], we have $f(x, 0) = 0$ , $f(0, \psi) \leq 0$ and $f(x, \delta_{3}) \leq 0$ by taking $\mu_{1}$ large depending on $\mu_{2}$ . We claim that the maximum of $f$ cannot be achieved in the interior.

By (1.6), we have

$\begin{align*} \begin{aligned} \partial_{x} \partial_{x}^{k} w-&\sqrt{w} \partial_{\psi}^{2} \partial_{x}^{k} w\\ & = -2 \partial_{x}^{k}(\rho\partial_{x}P)+\sum\limits_{m = 0}^{k-1} C_{k}^{m}(\partial_{x}^{k-m} \sqrt{w}) \partial_{\psi}^{2} \partial_{x}^{m}w +2\sum\limits_{m = 0}^{k} C_{k}^{m}\partial_{x}^{k-m}\left(\frac{\partial_{x}\rho}{\rho}\right)\partial_{x}^{m}w, \end{aligned} \end{align*}$

and

$\begin{align*} \partial_{\psi}^{2} \partial_{x}^{m} w = \partial_{x}^{m} \partial_{\psi}^{2} w = \partial_{x}^{m}\left(\frac{\partial_{x} w}{\sqrt{w}}+\frac{2\rho\partial_{x}P}{\sqrt{w}}-\frac{2\partial_{x}\rho}{\rho}\sqrt{w} \right) . \end{align*}$

For any $x \geq \frac{7 \varepsilon}{8}$ , $0 \leq j \leq k-1$ and $0 \leq m \leq k-1$ , from (2.6) and (2.14), we get

$\begin{align*} |\partial_{x}^{j} w| \leq C \psi, \quad|\partial_{x}^{k} w| \leq C \psi^{1-\beta},\quad |\partial_{x}^{k-m} \sqrt{w}| \leq C \psi^{\frac{1}{2}-\beta}. \end{align*}$

Then let $\beta\ll\frac{1}{2}$ , for $0 \leq m \leq k-1$ and $x \geq \frac{7 \varepsilon}{8}$ , we obtain

$\begin{align*} |\partial_{\psi}^{2} \partial_{x}^{m} w| \leq C \psi^{\frac{1}{2}-\beta}+C \psi^{-\frac{1}{2}}+C \psi^{\frac{1}{2}-\beta}\leq C \psi^{-\frac{1}{2}}. \end{align*}$

Therefore, we conclude that for $x \geq \frac{7 \varepsilon}{8}$ ,

$\begin{align*} \partial_{x} \partial_{x}^{k} w-\sqrt{w} \partial_{\psi}^{2} \partial_{x}^{k} w \leq C+C \psi^{-\beta}+C\psi^{1-\beta}\leq C \psi^{-\beta}. \end{align*}$

By the above inequality and (2.1), it holds at $p = p_{0}$ that

$\begin{align*} \begin{aligned} \partial_{x} f-\sqrt{w} \partial_{\psi}^{2} f& = \partial_{x} \partial_{x}^{k} w-\sqrt{w} \partial_{\psi}^{2} \partial_{x}^{k} w+\partial_{x}^{k}w\partial_{x}\phi-\sqrt{w} \partial_{\psi}^{2}(-\mu_{1} \psi+\mu_{2} \psi^{\frac{3}{2}-\beta})\\& \leq C_{2} \psi^{-\beta}-\xi \mu_{2} \psi^{-\beta}, \end{aligned} \end{align*}$

where $\xi = \left(\frac{3}{2}-\beta\right)\left(\frac{1}{2}-\beta\right) \sqrt{m} > 0$ . Then we have $\partial_{x} f-\sqrt{w} \partial_{\psi}^{2} f < 0$ in $\Omega$ by taking $\mu_{2}$ large depending on $C_{2}$ . This means that the maximum of $f$ cannot be achieved in the interior. Therefore, we have

$\max \limits_{\bar{\Omega}} f \leq 0.$

In the same way, we can prove that

$\max\limits _{\bar{\Omega}}-\partial_{x}^{k} w \phi-\mu_{1} \psi+\mu_{2} \psi^{\frac{3}{2}-\beta} \leq 0.$

So, for any $(x, \psi)\in[\frac{15}{16} \varepsilon, X] \times[0, \delta_{3}]$ , we have

$\begin{align*} |\partial_{x}^{k} w| \leq \mu_{1} \psi-\mu_{2} \psi^{\frac{3}{2}-\beta} \leq \mu_{1} \psi. \end{align*}$

Combining with (2.6) and (2.13), it holds in $[\frac{15}{16} \varepsilon, X] \times[0, \delta_{3}]$ that

$\begin{align*} |\partial_{x}^{k} \sqrt{w}| \leq C \psi^{\frac{1}{2}}. \end{align*}$

This completes the proof of the lemma.

Lemma 2.6. If $u$ is a solution for equation (1.1) in Theorem 1.1, assume $u_{0}$ satisfies the condition (1.9) and the known function $\rho$ and $\partial_{x}P$ are smooth. Assume that (2.6) holds, then for any $(x, \psi)\in[\varepsilon, X] \times[0, \delta_{4}]$ ,

$\begin{align*} |\partial_{\psi} \partial_{x}^{k} w| \leq C, \quad|\partial_{\psi}^{2} \partial_{x}^{k} w| \leq C \psi^{-1}, \quad|\partial_{x}^{k+1} w| \leq C \psi^{-\frac{1}{2}}. \end{align*}$

Proof. By Lemma 2.5 and (2.6), for any $(x, \psi)\in[\frac{15}{16} \varepsilon, X] \times[0, \delta_{3}]$ ,

$\begin{align} |\partial_{x}^{j} w| \leq C \psi, \quad|\partial_{x}^{j} \sqrt{w}| \leq C \psi^{\frac{1}{2}},\quad 0 \leq j \leq k. \end{align}$

(2.15)

Set $\Psi_{0} = \min \{\frac{2}{3} \delta_{3}, \frac{\varepsilon}{16}\}$ , for $(x_{0}, \psi_{0}) \in[\varepsilon, X] \times(0, \Psi_{0}]$ , we denote

$\begin{align*} \Omega = \bigl\{(x,\psi)|x_{0}-\psi_{0}^{\frac{3}{2}} \leq x \leq x_{0},\frac{1}{2} \psi_{0} \leq \psi \leq \frac{3}{2} \psi_{0}\bigr\}. \end{align*}$

A direct calculation gives

$\begin{align*} \begin{aligned} \partial_{x} \partial_{x}^{k} w-\sqrt{w} \partial_{\psi}^{2} \partial_{x}^{k} w = &-2 \partial_{x}^{k}(\rho\partial_{x}P)+\partial_{x}^{k} \sqrt{w} \partial_{\psi}^{2} w+\sum\limits_{m = 1}^{k-2} C_{k}^{m}\left(\partial_{x}^{k-m} \sqrt{w}\right) \partial_{\psi}^{2} \partial_{x}^{m} w \\ &+C_{k}^{k-1} \frac{\partial_{x} w}{2 \sqrt{w}} \partial_{\psi}^{2} \partial_{x}^{k-1} w+2\sum\limits_{m = 0}^{k} C_{k}^{m}\partial_{x}^{k-m}\left(\frac{\partial_{x}\rho}{\rho}\right)\partial_{x}^{m}w. \end{aligned} \end{align*}$

By (1.6), we obtain

$\begin{align*} \begin{aligned} \partial_{\psi}^{2} \partial_{x}^{m} w & = \partial_{x}^{m} \partial_{\psi}^{2} w\\ & = \partial_{x}^{m}\left(\frac{\partial_{x} w}{\sqrt{w}}+\frac{2\rho\partial_{x}P}{\sqrt{w}}-2\frac{\partial_{x}\rho}{\rho}\sqrt{w}\right) \\ & = \frac{\partial_{x}^{m+1} w}{\sqrt{w}}+\sum\limits_{l = 1}^{m} C_{m}^{l} \partial_{x}^{m-l+1} w \partial_{x}^{l} \frac{1}{\sqrt{w}}+\partial_{x}^{m}\left(\frac{2\rho\partial_{x}P}{\sqrt{w}}\right)-\partial_{x}^{m}\left(2\frac{\partial_{x}\rho}{\rho}\sqrt{w}\right), \end{aligned} \end{align*}$

and

$\begin{align*} \partial_{x}^{k} \sqrt{w} = \partial_{x}^{k-1} \frac{\partial_{x} w}{2 \sqrt{w}} = \frac{\partial_{x}^{k} w}{2 \sqrt{w}}+\sum\limits_{l = 1}^{k-1} C_{k-1}^{l} \partial_{x}^{k-1-l+1} w \partial_{x}^{l} \frac{1}{2 \sqrt{w}}, \end{align*}$

then

$\begin{align*} \begin{aligned} \partial_{x} \partial_{x}^{k} w-&\sqrt{w} \partial_{\psi}^{2} \partial_{x}^{k} w\\ = &-2 \partial_{x}^{k}(\rho\partial_{x}P)+\sum\limits_{m = 1}^{k-2} C_{k}^{m} (\partial_{x}^{k-m} \sqrt{w}) \partial_{\psi}^{2} \partial_{x}^{m} w \\ &+\frac{\partial_{x}^{k} w}{2 \sqrt{w}} \partial_{\psi}^{2} w+\sum\limits_{l = 1}^{k-1} C_{k-1}^{l} \partial_{x}^{k-1-l+1} w\partial_{x}^{l} \left(\frac{1}{2 \sqrt{w}}\right) \partial_{\psi}^{2} w \\ &+C_{k}^{k-1} \frac{\partial_{x} w \partial_{x}^{k} w}{2 w}+2\sum\limits_{m = 0}^{k} C_{k}^{m}\partial_{x}^{k-m}\left(\frac{\partial_{x}\rho}{\rho}\right)\partial_{x}^{m}w\\ &+C_{k}^{k-1} \frac{\partial_{x} w}{2 \sqrt{w}}\left[\sum\limits_{l = 1}^{k-1} C_{k-1}^{l} \partial_{x}^{k-l} w \partial_{x}^{l} \frac{1}{\sqrt{w}}+\partial_{x}^{k-1}\left(\frac{2\rho\partial_{x}P}{\sqrt{w}} \right)-\partial_{x}^{k-1}\left(2\frac{\partial_{x}\rho}{\rho}\sqrt{w}\right)\right]. \end{aligned} \end{align*}$

The following transformation $f$ is defined:

where $x-x_{0} = \psi_{0}^{\frac{3}{2}} \tilde{x}, \psi-\psi_{0} = \psi_{0} \tilde{\psi}$ .

Let $f = \partial_{x}^{k} w \psi_{0}^{-1}$ , we get

$\begin{align*} \begin{aligned} \partial_{\tilde{x}} f-&\frac{\sqrt{w}}{\psi_{0}^{\frac{1}{2}}} \partial_{\tilde{\psi}}^{2} f-\frac{1}{2 \sqrt{w}} \partial_{\psi}^{2} w \psi_{0}^{\frac{3}{2}}f-\frac{\partial_{x} w}{2 w} \psi_{0}^{\frac{3}{2}} f\\ & = -2\psi_{0}^{\frac{1}{2}} \partial_{x}^{k}(\rho\partial_{x}P)+\psi_{0}^{\frac{1}{2}}\sum\limits_{m = 1}^{k-2} C_{k}^{m} (\partial_{x}^{k-m} \sqrt{w}) \partial_{\psi}^{2} \partial_{x}^{m} w \\ &\quad+\psi_{0}^{\frac{1}{2}}\sum\limits_{l = 1}^{k-1} C_{k-1}^{l} \partial_{x}^{k-l} w\left(\partial_{x}^{l} \frac{1}{2 \sqrt{w}}\right) \partial_{\psi}^{2} w \\ &\quad+2\psi_{0}^{\frac{1}{2}}\sum\limits_{m = 0}^{k} C_{k}^{m}\partial_{x}^{k-m}\left(\frac{\partial_{x}\rho}{\rho}\right)\partial_{x}^{m}w\\ &\quad+\psi_{0}^{\frac{1}{2}}\frac{\partial_{x} w}{2 \sqrt{w}}\left[\sum\limits_{l = 1}^{k-1} C_{k-1}^{l} \partial_{x}^{k-l} w \partial_{x}^{l} \frac{1}{\sqrt{w}}+\partial_{x}^{k-1}\left(\frac{2\rho\partial_{x}P}{\sqrt{w}} \right)-\partial_{x}^{k-1}\left(2\frac{\partial_{x}\rho}{\rho}\sqrt{w}\right)\right]\\ &: = F. \end{aligned} \end{align*}$

From the proof of Lemma 2.2 and Lemma 2.6, we know that in $\widetilde{\Omega}$ for $\alpha \in(0, 1)$ ,

$\begin{align*} |f|\leq C, \quad 0 < c \leq \psi_{0}^{-\frac{1}{2}}\sqrt{w} \leq C, \quad|\psi_{0}^{-\frac{1}{2}}\sqrt{w}|_{\mathcal{C}^{\alpha}(\widetilde{\Omega})} \leq C . \end{align*}$

By (2.6), (2.15) and the equality

$\begin{align*} \begin{aligned} \partial_{\psi}\left(\partial_{\psi}^{2} \partial_{x}^{m} w\right) = & \frac{\partial_{\psi} \partial_{x}^{m+1} w}{\sqrt{w}}-\frac{\partial_{\psi} w \partial_{x}^{m+1} w}{2(\sqrt{w})^{3}}+\sum\limits_{l = 1}^{m} C_{m}^{l} \partial_{x}^{m-l+1} \partial_{\psi} w \partial_{x}^{l} \frac{1}{\sqrt{w}} \\ &+\sum\limits_{l = 1}^{m} C_{m}^{l} \partial_{x}^{m-l+1} w \partial_{x}^{l} \frac{\partial_{\psi} w}{-2(\sqrt{w})^{3}}+\partial_{x}^{m}\left(\frac{\rho\partial_{x}P\partial_{\psi} w}{-(\sqrt{w})^{3}} \right)-\partial_{x}^{m}\left(\frac{\partial_{x}\rho}{\rho}\frac{\partial_{\psi}w}{\sqrt{w}}\right), \end{aligned} \end{align*}$

we can conclude that for $j \leq k-1$ and $m \leq k-2$ ,

$\begin{align*} \big|\nabla_{\tilde{x}, \tilde{\psi}} \partial_{x}^{j} \sqrt{w}\big| \leq C \psi_{0}^{\frac{1}{2}}, \quad\big|\nabla_{\tilde{x}, \tilde{\psi}} \partial_{x}^{j}\big(\frac{1}{\sqrt{w}}\big)\big| \leq C \psi_{0}^{-\frac{1}{2}}, \quad\big|\nabla_{\tilde{x}, \tilde{\psi}} \partial_{\psi}^{2} \partial_{x}^{m} w\big | \leq C \psi_{0}^{-\frac{1}{2}}. \end{align*}$

Combining (2.4) with (2.5), we can obtain

$\begin{align*} |\frac{1}{2 \sqrt{w}} \partial_{\psi}^{2} w \psi_{0}^{\frac{3}{2}}+\frac{\partial_{x} w}{2 w} \psi_{0}^{\frac{3}{2}}|_{\mathcal{C}^{\alpha}(\widetilde{\Omega})}+|F|_{\mathcal{C}^{\alpha}(\widetilde{\Omega})} \leq C . \end{align*}$

By the standard interior priori estimates, we obtain

Therefore, this means that

$\begin{align*} |\partial_{x}^{k+1} w\left(x_{0}, \psi_{0}\right)| \leq C \psi_{0}^{-\frac{1}{2}}, \quad|\partial_{\psi} \partial_{x}^{k} w\left(x_{0}, \psi_{0}\right)| \leq C, \quad|\partial_{\psi}^{2} \partial_{x}^{k} w\left(x_{0}, \psi_{0}\right)| \leq C \psi_{0}^{-1} . \end{align*}$

Since $\left(x_{0}, \psi_{0}\right)$ is arbitrary, this completes the proof of the lemma.

Lemma 2.7. If $u$ is a solution for equation (1.1) in Theorem 1.1, assume $u_{0}$ satisfies the condition (1.9) and the known function $\rho$ and $\partial_{x}P$ are smooth. Assume $0 < \varepsilon < X$ and integer $m, k \geq 0$ , then there exists a positive constant $\delta > 0$ such that for any $(x, \psi)\in[\varepsilon, X]\times[0, \delta],$

$\begin{align} |\partial_{\psi}^{m} \partial_{x}^{k} w| \leq C \psi^{1-m}. \end{align}$

(2.16)

Proof. From Lemma 2.1, (2.1), Lemma 2.2 and Lemma 2.3, a direct calculation can prove that

$\begin{align*} \big|\partial_{x}^{k} \frac{1}{\sqrt{w}}\big| \leq C \psi^{-\frac{1}{2}}, \quad\big|\partial_{x}^{k} \partial_{\psi} \frac{1}{\sqrt{w}}\big| \leq C \psi^{-\frac{3}{2}}, \quad\big|\partial_{x}^{k} \partial_{\psi}^{2} \frac{1}{\sqrt{w}}\big| \leq C \psi^{-\frac{5}{2}} , \end{align*}$

and (2.16) holds for $m = 0, 1, 2$ . Then for $0 \leq m \leq j$ with $j \geq 1$ , we inductively assume that

$\begin{align} |\partial_{\psi}^{m} \partial_{x}^{k} w| \leq C \psi^{1-m},\quad \big|\partial_{x}^{k} \partial_{\psi}^{m} \frac{1}{\sqrt{w}}\big| \leq C \psi^{-\frac{1}{2}-m}. \end{align}$

(2.17)

In the next part, we will prove that (2.17) still holds for $m = j+1$ .

By (1.6), we obtain

$\begin{align*} \begin{aligned} \partial_{\psi}^{j+1} \partial_{x}^{k} w & = \partial_{\psi}^{j-1} \partial_{x}^{k} \partial_{\psi}^{2} w\\& = \partial_{x}^{k} \partial_{\psi}^{j-1}\left(\frac{\partial_{x} w}{\sqrt{w}}+\frac{2\rho\partial_{x}P}{\sqrt{w}} -2\frac{\partial_{x}\rho}{\rho}\frac{w}{\sqrt{w}}\right) \\ & = \partial_{x}^{k}\left(\sum\limits_{i = 0}^{j-1} C_{j-1}^{i} \partial_{\psi}^{j-1-i} \partial_{x} w \partial_{\psi}^{i} \frac{1}{\sqrt{w}}+2 \rho\partial_{x}P \partial_{\psi}^{j-1} \frac{1}{\sqrt{w}}-2\frac{\partial_{x}\rho}{\rho} \sum\limits_{i = 0}^{j-1} C_{j-1}^{i} \partial_{\psi}^{j-1-i} w \partial_{\psi}^{i} \frac{1}{\sqrt{w}}\right). \end{aligned} \end{align*}$

Combining with (2.17), we get

$\begin{align} |\partial_{\psi}^{j+1} \partial_{x}^{k} w| \leq C \psi^{\frac{3}{2}-j}+ C \psi^{\frac{1}{2}-j}+ C \psi^{\frac{3}{2}-j}\leq C \psi^{\frac{1}{2}-j} . \end{align}$

(2.18)

By straight calculations, we get

$\begin{align*} \begin{aligned} 0& = \partial_{x}^{k} \partial_{\psi}^{j+1}\bigg(\frac{1}{\sqrt{w}} \frac{1}{\sqrt{w}} w\bigg) \\ & = \partial_{x}^{k}\bigg[2 \sqrt{w} \partial_{\psi}^{j+1} \frac{1}{\sqrt{w}}+\sum\limits_{i = 1}^{j} \sum\limits_{l = 0}^{j+1-i} C_{j+1}^{i} C_{j+1-i}^{l}\left(\partial_{\psi}^{i} \frac{1}{\sqrt{w}}\right)\left(\partial_{\psi}^{l} \frac{1}{\sqrt{w}}\right) \partial_{\psi}^{j+1-l-i} w\\ &\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\quad +\sum\limits_{l = 0}^{j} C_{j+1}^{l} \frac{1}{\sqrt{w}}\left(\partial_{\psi}^{l} \frac{1}{\sqrt{w}}\right) \partial_{\psi}^{j+1-l} w\bigg]. \end{aligned} \end{align*}$

Combining the above equality with (2.17), we can conclude that

$\begin{align*} \big|\partial_{x}^{k} \partial_{\psi}^{j+1} \frac{1}{\sqrt{w}}\big| \leq C \psi^{-\frac{3}{2}-j}. \end{align*}$

This completes the proof of the lemma.

3. Proof of Theorem 1.5

In this section, we will prove the regularity of the solution $u$ in the domain

$\left\{(x,\psi)|\varepsilon\leq x\leq X, 0\leq y \leq Y_{1}\right\}.$

Proof of Theorem 1.5:

Proof. For the convenience of proof, we denote

$\begin{align*} (\tilde{x}, \psi) = \left(x, \int_{0}^{y} \tilde{u} d y\right) . \end{align*}$

A direct calculation gives (see P13 in ^[23])

$\begin{align*} \begin{aligned} \partial_{y} = \sqrt{w} \partial_{\psi}, \quad \partial_{x} = \partial_{\tilde{x}}+\partial_{x} \psi(x, y) \partial_{\psi}, \quad \partial_{x} \psi = \frac{1}{2} \sqrt{w} \int_{0}^{\psi} w^{-\frac{3}{2}} \partial_{\tilde{x}} w d \psi . \end{aligned} \end{align*}$

By (2.1) and Lemma 2.3, we have $|\partial_{x} \psi| \leq C \psi$ . Due to $\partial_{y} = \sqrt{w} \partial_{\psi}$ , we obtain

$\begin{align*} \begin{aligned} \partial_{x}^{k} 2 \partial_{y} \tilde{u} & = \left(\partial_{\tilde{x}}+\partial_{x} \psi \partial_{\psi}\right)^{k} \partial_{\psi} w, \\ \partial_{x}^{k} 2 \partial_{y}^{2} \tilde{u} & = \left(\partial_{\tilde{x}}+\partial_{x} \psi \partial_{\psi}\right)^{k}\left(\partial_{\tilde{x}} w+2 \rho\partial_{x}P-2\frac{\partial_{x}\rho}{\rho}w\right) \\ & = \left(\partial_{\tilde{x}}+\partial_{x} \psi \partial_{\psi}\right)^{k}\left(\partial_{\tilde{x}} w\right)+2 \partial_{\tilde{x}}^{k}(\rho\partial_{x}P)-2\left(\frac{\partial_{x}\rho}{\rho}\right)(\partial_{\tilde{x}}+\partial_{x} \psi \partial_{\psi})^{k}w-2\partial_{\tilde{x}}^{k}\left(\frac{\partial_{x}\rho}{\rho}\right)w. \end{aligned} \end{align*}$

By $|\partial_{x} \psi| \leq C \psi$ and Lemma 2.7, we obtain that Theorem 1.5 holds for $m = 0, 1, 2,$

$\begin{align} |\partial_{x}^{k} \partial_{y} \tilde{u}|+|\partial_{x}^{k} \partial_{y}^{2} \tilde{u}| \leq C . \end{align}$

(3.1)

We inductively assume that for any integer $k$ and $m \geq 1$ ,

$\begin{align} |\partial_{x}^{k} \partial_{y}^{j}\tilde{u}| \leq C, \quad j \leq m . \end{align}$

(3.2)

A direct calculation gives

$\begin{align*} \begin{aligned} &\partial_{x}^{k} \partial_{y}^{m+1} \tilde{u} \\ & = \partial_{x}^{k} \partial_{y}^{m-1} \partial_{y}^{2} \tilde{u}\\ & = \partial_{x}^{k} \partial_{y}^{m-1}\left(\tilde{u} \partial_{x} \tilde{u}-\partial_{y}\tilde{u} \int_{0}^{y}\partial_{x} \tilde{u}d y-\frac{\partial_{x}\rho}{\rho}\tilde{u}^{2}\right) \\ & = \partial_{x}^{k}\bigg(\sum\limits_{i = 0}^{m-1} C_{m-1}^{i} \partial_{y}^{m-1-i}\tilde{u}\partial_{y}^{i} \partial_{x} \tilde{u}-\sum\limits_{i = 0}^{m-2} C_{m-1}^{i+1} \partial_{y}^{m-1-i}\tilde{u} \partial_{y}^{i} \partial_{x} \tilde{u}-\partial_{y}^{m}\tilde{u} \int_{0}^{y}\partial_{x} \tilde{u}d y-\frac{\partial_{x}\rho}{\rho}\partial_{y}^{m-1}\tilde{u}^{2}\bigg), \end{aligned} \end{align*}$

and we can deduce from (3.1) and (3.2) that

$\begin{align*} |\partial_{x}^{k} \partial_{y}^{j} \tilde{u}| \leq C, \quad j \leq m+1 .\quad \Rightarrow \quad |\partial_{x}^{k} \partial_{y}^{j} u| \leq C, \quad j \leq m+1. \end{align*}$

This completes the proof of the theorem.

4. Proof of Theorem 1.3 and 1.6

In this section, we prove our main theorem. The key point is to prove that (1.6) is a uniform parabolic equation. The proof is based on the classical parabolic maximum principle. The specific proof details are as follows.

Proof. By (1.2) and $\partial_{x}P \leq 0$ , we obtain

$\begin{align*} C \geq U^{2}(x) = U^{2}(0)-2 \int_{0}^{x} \frac{\partial_{x}P(\rho)}{\rho} d x \geq U^{2}(0). \end{align*}$

By (1.7) and $w$ increasing in $\psi$ (see below), we know that there exists some positive constants $\Psi$ and $C_{0}$ such that for any $(x, \psi)\in[0, X] \times[\Psi, +\infty),$

$\begin{align} w \geq C_{0}U^{2}(0). \end{align}$

(4.1)

From Theorem 1.1, we know that there exists positive constants $y_{0}, M, m$ such that for any $(x, \psi)\in[0, X] \times[0, y_{0}]$ (we can take $y_{0}$ to be small enough),

$\begin{align} M \geq \partial_{y} \tilde{u}(x, y) \geq m . \end{align}$

(4.2)

The fact that $\psi\sim y^{2}$ is near the boundary $y = 0$ (see Remark 4.1 in ^[23]), for some small positive constant $0 < \kappa < 1$ , we get

$\begin{align} \frac{\kappa}{2} y_{0}^{2} \leq \psi \leq \kappa y_{0}^{2} \Rightarrow \sigma y_{0} \leq y \leq \frac{y_{0}}{2}, \end{align}$

(4.3)

for some constant $\sigma > 0$ depends on $\kappa, m, M$ .

We denote

$\begin{align*} \Omega = \bigl\{(x,\psi)|0\leq x\leq X,\frac{\kappa}{2} y_{0}^{2}\leq \psi\leq +\infty\bigr\}. \end{align*}$

By (4.2) and (4.3), we get $\tilde{u}(x, \sigma y_{0}) \geq m \sigma y_{0}$ , then for any $x\in[0, X]$ , we have

$\begin{align} w(x, \frac{\kappa}{2} y_{0}^{2}) \geq m^{2} \sigma^{2} y_{0}^{2}. \end{align}$

(4.4)

Since the initial data $u_{0}$ satisfies the condition (1.9) and $w = \tilde{u}^{2}$ , we know $w(0, \psi) > 0$ for $\psi > 0$ and there exists a positive constant $\zeta$ , such that for $\psi \in[\frac{\kappa}{2} y_{0}^{2}, \Psi]$ ,

$\begin{align} w(0, \psi) > \zeta. \end{align}$

(4.5)

Then, we only consider

$\begin{align*} \Omega_{1} = \bigl\{(x, \psi)|0\leq x\leq X, \frac{\kappa}{2} y_{0}^{2}\leq\psi\leq \Psi\bigr\}. \end{align*}$

We denote $H(x, \psi): = e^{-\lambda x}\partial_{\psi}w(x, \psi)$ , which satisfies the following system in the region $\Omega_{0} = \left\{(x, \psi)|0\leq x < X, 0 < \psi < +\infty\right\}$ :

$\begin{align} \left\{\begin{aligned} &\partial_{x}H-\frac{\partial_{\psi}w}{2\sqrt{w}}\partial_{\psi}H-\sqrt{w}\partial_{\psi}^{2}H+(\lambda-2\frac{\partial_{x}\rho}{\rho})H = 0,\\ &H|_{x = 0} = \partial_{\psi}w_{0}(\psi),\quad H|_{\psi = 0} = 2e^{-\lambda x}\partial_{y}\tilde{u}|_{y = 0},\quad H|_{\psi = +\infty} = 0. \end{aligned}\right. \end{align}$

(4.6)

Then, we choose $\lambda$ properly large such that $\lambda-2\frac{\partial_{x}\rho}{\rho}\geq 0$ . Due to

$\begin{align*} H|_{x = 0} = \partial_{\psi}w_{0}(\psi)\geq0,\quad H|_{\psi = 0} = 2e^{-\lambda x}\partial_{y}\tilde{u}|_{y = 0} > 0,\quad H|_{\psi = +\infty} = 0, \end{align*}$

it follows that

$\begin{align*} H(x,\psi) = e^{-\lambda x}F(x,\psi) = e^{-\lambda x}\partial_{\psi}w\geq0,\quad (x, \psi)\in[0,X^{*})\times\mathbb{R}_{+}, \end{align*}$

which means $\partial_{\psi}w\geq0$ in $[0, X)\times\mathbb{R}_{+}$ . Hence, $w$ is increasing in $\psi$ . Therefore, we know that there exists a positive constant $\lambda\geq m^{2} \sigma^{2} y_{0}^{2}$ such that for any $x\in[0, X],$

$\begin{align} w(x, \Psi) \geq \lambda. \end{align}$

(4.7)

By (1.6), for any $\varepsilon > 0$ , we know $W: = w+\varepsilon x$ satisfies the following system in $\Omega_{1}:$

$\begin{align*} \left\{\begin{aligned} &\partial_{x}W-\sqrt{w}\partial_{\psi}^{2}W-2\frac{\partial_{x}\rho}{\rho}W = \mathcal{F},\\ &W|_{x = 0} = W_{0} > \zeta,\quad W|_{\psi = \frac{\kappa}{2}y^{2}_{0}} = W_{1}\geq m^{2}\sigma^{2}y^{2}_{0},\quad W|_{\psi = \Psi} = W_{2}\geq \lambda, \end{aligned}\right. \end{align*}$

where

$\begin{align*} \mathcal{F} = -2\rho\partial_{x}P+\varepsilon-2\varepsilon x \frac{\partial_{x}\rho}{\rho}. \end{align*}$

Since $\partial_{x}P\leq0$ , we know the diffusive term $\mathcal{F} > 0.$ Therefore, the minimum cannot be reached inside $\Omega_{1}$ . Set

$\begin{align*} \eta_{0} = \min\left\{W_{0}, W_{1}, W_{2}\right\}, \end{align*}$

then by the maximum principle, we obtain $W = w+\varepsilon x\geq \eta_{0}$ . Let $\varepsilon \rightarrow 0$ , we have $w \geq \eta_{0}$ in $\Omega_{1}$ . Then we denote

$\begin{align*} \eta = \min\left\{\eta_{0}, C_{0}U^{2}(0)\right\} > 0, \end{align*}$

combining with (4.1), we have $w\geq\eta$ in $\Omega$ . Therefore, there exists some positive constant $c$ such that $c\leq w$ in $\Omega$ . From Theorem 1.1, we have $w \leq C$ in $\Omega$ . In sum, there exists positive constants $c, C$ such that $c \leq w \leq C$ in $\Omega$ . This further means that

$\begin{align} 0 < \sqrt{c}\leq\sqrt{w}\leq\sqrt{C}, \end{align}$

(4.8)

where $C$ depends on $X$ . Therefore, we prove (1.6) is a uniform parabolic equation. Furthermore, by Theorem 1.1, we know $\partial_{y}\tilde{u}, \partial_{y}^{2}\tilde{u}$ are continuous and bounded in $[0, X)\times \mathbb{R}_{+}$ . Combining $\rho$ , $\partial_{x}P$ are smooth, (4.8) with

$\begin{align*} 2\partial_{y}\tilde{u} = \partial_{\psi}w,\quad 2\partial_{y}^{2}\tilde{u} = \sqrt{w}\partial_{\psi}^{2}w = \partial_{x}w-2\frac{\partial_{x}\rho}{\rho}w+2\rho\partial_{x}P(\rho), \end{align*}$

we obtain

$\begin{align*} \|\sqrt{w}\|_{\mathcal{C}^{\alpha}(\Omega)}\leq C. \end{align*}$

Once we have the above conclusion, the proof of Theorem 1.6 can be given in a similar fashion to ^[23]. Here, we provide a brief explanation for the reader's convenience. More details can be found in ^[23].

Step 1: For any $(x_{1}, \psi_{1})\in[\varepsilon, X]\times[ \kappa y_{0}^{2}, +\infty)$ , we denote

$\begin{align*} \Omega_{x_{1}, \psi_{1}} = \bigl\{(x, \psi)|x_{1}-\frac{\varepsilon}{2}\leq x\leq x_{1}, \psi_{1}-\frac{\kappa}{2} y_{0}^{2}\leq\psi\leq \psi_{1}+\frac{\kappa}{2} y_{0}^{2}\bigr\}. \end{align*}$

Step 2: Note that the known function $\rho$ , $\partial_{x}P$ is smooth, we can repeat interior Schauder estimates in $\Omega_{x_{1}, \psi_{1}}$ to achieve uniform estimates independent of choice of $(x_{1}, \psi_{1})$ for any order derivatives of $w$ . Since the width and the length of $\Omega_{x_{1}, \psi_{1}}$ are constants and the estimates employed are independent of $(x_{1}, \psi_{1})$ , restricting the estimates to the point $(x_{1}, \psi_{1})$ , we can get for any $m < +\infty, |\nabla^{m}w(x_{1}, \psi)|\leq C_{X, m, y_{0}, \varepsilon}$ .

Step 3: Since $(x_{1}, \psi_{1})$ is arbitrary, we have for any $m < +\infty, |\nabla^{m}w(x_{1}, \psi)|\leq C_{X, m, y_{0}, \varepsilon}$ in $[\varepsilon, X]\times[ \kappa y_{0}^{2}, +\infty)$ . Then, as in Section 3, we can prove Theorem 1.6.

Finally, Theorem 1.3 is proven by combining Theorem 1.5 and Theorem 1.6.

Use of AI tools declaration

The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.

Acknowledgments

The research of Zou was supported by the Fundamental Research Funds for the Central Universities (Grant No. 202261101).

Conflict of interest

The authors declare there is no conflict of interest.

References

[1]	J. Chen, Y. Lu, Q. Yu, X. Luo, E. Adeli, Y. Wang, et al., TransUNet: Transformers make strong encoders for medical image segmentation, preprint, arXiv: 2102.04306.
[2]	T. P. Shen, H. Q. Xu, Medical image segmentation based on Transformer and HarDNet structures, IEEE Access, 11 (2023), 16621–16630. https://doi.org/10.1109/ACCESS.2023.3244197 doi: 10.1109/ACCESS.2023.3244197
[3]	L. Han, Y. H. Chen, J. M. Li, B. W. Zhong, Y. Z. Lei, M. H. Sun, Liver segmentation with 2.5 D perpendicular UNets, Comput. Electr. Eng., 91 (2021), 107118. https://doi.org/10.1016/j.compeleceng.2021.107118 doi: 10.1016/j.compeleceng.2021.107118
[4]	H. Y. Li, X. Q. Zhao, A. Y. Su, H. T. Zhang, J. X. Liu, G. Y. Gu, Color space transformation and multi-class weighted loss for adhesive white blood cell segmentation, IEEE Access, 8 (2020), 24808–24818. https://doi.org/10.1109/ACCESS.2020.2970485 doi: 10.1109/ACCESS.2020.2970485
[5]	T. Magadza, S. Viriri, Deep learning for brain tumor segmentation: a survey of state-of-the-art, J. Imaging, 7 (2021), 19. https://doi.org/10.3390/jimaging7020019 doi: 10.3390/jimaging7020019
[6]	Y. E. Almalki, A. Qayyum, M. Irfan, N. Haider, A. Glowacz, F. M. Alshehri, et al., A novel method for COVID-19 diagnosis using artificial intelligence in chest X-ray images, Healthcare, 9 (2021), 522. https://doi.org/10.3390/healthcare9050522 doi: 10.3390/healthcare9050522
[7]	D. Q. Zhang, S. C. Chen, A novel kernelized fuzzy c-means algorithm with application in medical image segmentation, Artif. Intell. Med., 32 (2014), 37–50. https://doi.org/10.1016/j.artmed.2004.01.012 doi: 10.1016/j.artmed.2004.01.012
[8]	H. P. Ng, S. H. Ong, K. W. C. Foong, Poh-Sun Goh, W. L. Nowinski, Medical image segmentation using k-means clustering and improved watershed algorithm, in 2006 IEEE southwest symposium on image analysis and interpretation, (2006). https://doi.org/10.1109/SSIAI.2006.1633722
[9]	N. A. Mohamed, M. N. Ahmed, A. Farag, Modified fuzzy c-mean in medical image segmentation, in 1999 IEEE International Conference on Acoustics, (1999). https://doi.org/10.1109/ICASSP.1999.757579
[10]	A. Prabin A, J. Veerappan, Automatic segmentation of lung ct images by CC based region growing, J. Theor. Appl. Inf. Technol., 68 (2014), 63–69.
[11]	M. Negassi, R. Suarez-Ibarrola, S. Hein, A. Miernik, A. Reiterer, Application of artificial neural networks for automated analysis of cystoscopic images: a review of the current status and future prospects, World J. Urol., 38 (2020), 2349–2358. https://doi.org/10.1007/s00345-019-03059-0 doi: 10.1007/s00345-019-03059-0
[12]	Y. Zhang, M. A. Khan, Z. Zhu, S. Wang, SNELM: SqueezeNet-guided ELM for COVID-19 recognition, Comput. Syst. Sci. Eng., 46 (2023), 13–26. https://doi.org/10.32604/csse.2023.034172 doi: 10.32604/csse.2023.034172
[13]	M. Irfan, M. A. Iftikhar, S. Yasin, U. Draz, T. Ali, S. Hussain, et al., Role of hybrid deep neural networks (HDNNs), computed tomography, and chest X-rays for the detection of COVID-19, Int. J. Environ. Res. Public Health, 18 (2021), 3056. https://doi.org/10.3390/ijerph18063056 doi: 10.3390/ijerph18063056
[14]	J. Long, E. Shelhamer, T. Darrell, Fully convolutional networks for semantic segmentation, in Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition, (2015). https://doi.org/10.1109/CVPR.2015.7298965
[15]	O. Ronneberger, P. Fischer, T. Brox, U-net: Convolutional networks for biomedical image segmentation, in International Conference on Medical image computing and computer-assisted intervention, (2015). https://doi.org/10.1007/978-3-319-24574-4_28
[16]	Z. W. Zhou, M. M. R. Siddiquee, N. Tajbakhsh, J. M. Liang, Unet++: redesigning skip connections to exploit multiscale features in image segmentation, IEEE Trans. Med. Imaging, 39 (2020), 1856–1867. https://doi.org/10.1109/TMI.2019.2959609 doi: 10.1109/TMI.2019.2959609
[17]	O. Oktay, J. Schlemper, L. L. Folgoc, M. Lee, M. Heinrich, K. Misawa, et al., Attention U-Net: learning where to look for the Pancreas, preprint, arXiv: 1804.03999.
[18]	D. L. Peng, S. Y. Xiong, W. J. Peng, J. P. Lu, LCP-net: a local context-perception deep neural network for medical image segmentation, Expert Syst. Appl., 168 (2021), 114234. https://doi.org/10.1016/j.eswa.2020.114234 doi: 10.1016/j.eswa.2020.114234
[19]	C. Chen, B. Liu, K. N. Zhou, W. Z. He, F. Yan, Z. L. Wang, R. X. Xiao, CSR-net: cross-scale residual network for multi- objective scaphoid fracture segmentation, Comput. Biol. Med., 137 (2021), 104776. https://doi.org/10.1016/j.compbiomed.2021.104776 doi: 10.1016/j.compbiomed.2021.104776
[20]	E. K. Wang, C. M. Chen, M. M. Hassan, A. Almogren, A deep learning based medical image segmentation technique in Internet-of-Medical-Things domain, Future Gene. Comput. Sy., 108 (2020), 135–144. https://doi.org/10.1016/j.future.2020.02.054 doi: 10.1016/j.future.2020.02.054
[21]	T. Feng, C. S. Wang, X. W. Chen, H. T. Fan, K. Zeng, Z. Y. Li, URNet: A UNet based residual network for image dehazing, Appl. Soft Comput., 102 (2020), 106884. https://doi.org/10.1016/j.asoc.2020.106884 doi: 10.1016/j.asoc.2020.106884
[22]	R. Q. Ge, H. H. Cai, X. Yuan, F. W. Qin, Y. Huang, et al., MD-UNET: Multiinput dilated U-shape neural network for segmentation of bladder cancer, Comput. Biol. Chem., 93 (2021), 107510. https://doi.org/10.1016/j.compbiolchem.2021.107510 doi: 10.1016/j.compbiolchem.2021.107510
[23]	Y. C. Lan, X. M. Zhang, Real-time ultrasound image despeckling using mixed-attention mechanism based residual UNet, IEEE Access, 8 (2020), 195327–195340. https://doi.org/10.1109/ACCESS.2020.3034230 doi: 10.1109/ACCESS.2020.3034230
[24]	C. Li, Y. S. Tan, W. Chen, X. Luo, Y. L. He, Y. M. Gao, F. Li, ANU-Net: Attention-based nested U-Net to exploit full resolution features for medical image segmentation, Comput. Graph, 90 (2020), 11–20. https://doi.org/10.1016/j.cag.2020.05.003 doi: 10.1016/j.cag.2020.05.003
[25]	C. L. Guo, M. Szemenyei, Y. G. Yi, W. L. Wang, B. Chen, C. Q. Fan, SA-UNet: Spatial attention U-Net for retinal vessel segmentation, in 25th International Conference on Pattern Recognition (ICPR), (2021). https://doi.org/10.1109/ICPR48806.2021.9413346
[26]	J. Bernal, F. J. Sánchez, G. Fernández-Esparrach, D. Gil, C. Rodríguez, F. Vilariño, WM-DOVA maps for accurate polyp highlighting in colonoscopy: validation vs. saliency maps from physicians, Comput. Med. Imaging Graphics, 43 (2015), 99–111. https://doi.org/10.1016/j.compmedimag.2015.02.007 doi: 10.1016/j.compmedimag.2015.02.007
[27]	J. Soltani-Nabipour, A. Khorshidi, B. Noorian, Lung tumor segmentation using improved region growing algorithm, Nuclear Eng. Technol., 52 (2020), 2313–2319. https://doi.org/10.1016/j.net.2020.03.011 doi: 10.1016/j.net.2020.03.011
[28]	S. Y. Chong, M. K. Tan, K. B. Yeo, M. Y. Ibrahim, X. Hao, K. T. K. Teo, Segmenting nodules of lung tomography image with level set algorithm and neural network, in 2019 IEEE 7th Conference on Systems, Process and Control (ICSPC), (2019). https://doi.org/10.1109/ICSPC47137.2019.9067987
[29]	M. Savic, Y. Ma, G. Ramponi, W. Du, Y. Peng, Lung nodule segmentation with a region-based fast marching method, Sensors, 21 (2021), 1908. https://doi.org/10.3390/s21051908 doi: 10.3390/s21051908
[30]	P. M. Bruntha, S. I. A. Pandian, P. Mohan, Active Contour Model (without edges) based pulmonary nodule detection in low dose CT images, in 2019 2nd International Conference on Signal Processing and Communication (ICSPC), (2019). https://doi.org/10.1109/ICSPC46172.2019.8976813
[31]	R. Manickavasagam, S. Selvan, GACM based segmentation method for Lung nodule detection and classification of stages using CT images, in 2019 1st International Conference on Innovations in Information and Communication Technology (ICIICT), (2019). https://doi.org/10.1109/ICIICT1.2019.8741477.
[32]	Y. LeCun, L. Bottou, Y. Bengio, P. Haffner, Gradient-based learning applied to document recognition, Proc. IEEE, 86 (1998), 2278–2324. https://doi.org/10.1109/5.726791 doi: 10.1109/5.726791
[33]	G. Simantiris, G. Tziritas, Cardiac MRI segmentation with a dilated CNN incorporating domain-specific constraints, IEEE J. Selected Topics Signal Process., 14 (2020), 1235–1243. https://doi.org/10.1109/JSTSP.2020.3013351 doi: 10.1109/JSTSP.2020.3013351
[34]	B. Thyreau, Y. Taki, Learning a cortical parcellation of the brain robust to the MRI segmentation with convolutional neural networks, Med. Image Anal., 14 (2020), 101639. https://doi.org/10.1016/j.media.2020.101639 doi: 10.1016/j.media.2020.101639
[35]	M. F. Aslan, A robust semantic lung segmentation study for CNN-based COVID-19 diagnosis, Chemom. Intell. Lab. Syst., 231 (2022), 104695. https://doi.org/10.1016/j.chemolab.2022.104695 doi: 10.1016/j.chemolab.2022.104695
[36]	S. Akila Agnes, J. Anitha, J. D. Peter, Automatic lung segmentation in low-dose chest CT scans using convolutional deep and wide network (CDWN), Neural Comput. Appl., 32 (2020), 15845-15855. https://doi.org/10.1007/s00521-018-3877-3 doi: 10.1007/s00521-018-3877-3
[37]	L. L. Du, H. R. Liu, L. Zhang, Y. Lu, M. Y. Li, Y. Hu, et al., Deep ensemble learning for accurate retinal vessel segmentation, Comput. Biol. Med., 158 (2023), 106829. https://doi.org/10.1016/j.compbiomed.2023.106829 doi: 10.1016/j.compbiomed.2023.106829
[38]	Y. Wu, L. Lin, Automatic lung segmentation in CT images using dilated convolution based weighted fully convolutional network, J. Phys. Confer. Ser., 1646 (2022), 012032. https://doi.org/10.1088/1742-6596/1646/1/012032 doi: 10.1088/1742-6596/1646/1/012032
[39]	H. Xia, W. Sun, S. Song, X. Mou, Md-net: multi-scale dilated convolution network for CT images segmentation, Neural Process. Lett., 51 (2020), 2915–2927. https://doi.org/10.1007/s11063-020-10230-x doi: 10.1007/s11063-020-10230-x
[40]	H. Liu, H. Cao, E. Song, G. Ma, X. Xu, R. Jin, C. C. Hung, A cascaded dual-pathway residual network for lung nodule segmentation in CT images, Phys. Med., 63 (2019), 112–121. https://doi.org/10.1016/j.ejmp.2019.06.003 doi: 10.1016/j.ejmp.2019.06.003
[41]	H. R. Roth, H. Oda, X. Zhou, N. Shimizu, Y. Yang, Y. Hayash, et al., An application of cascaded 3D fully convolutional networks for medical image segmentation, Comput. Med. Imaging Graphics, 66 (2018), 90–99. https://doi.org/10.1016/j.compmedimag.2018.03.001 doi: 10.1016/j.compmedimag.2018.03.001
[42]	A. Lin, B. Chen, J. Xu, Z. Zhang, G. Lu, D. Zhang, Ds-transunet: Dual swin transformer u-net for medical image segmentation, IEEE Trans. Instrum. Meas., 71 (2022), 1–15. https://doi.org/10.1109/TIM.2022.3178991 doi: 10.1109/TIM.2022.3178991
[43]	F. Milletari, N. Navab, S. A. Ahmadi, V-net: Fully convolutional neural networks for volumetric medical image segmentation, in 2016 Fourth International Conference on 3D Vision (3DV), (2016). https://doi.org/10.48550/arXiv.1606.04797
[44]	F. Hoorali, H. Khosravi, B. Moradi, IRUNet for medical image segmentation, Expert Syst. Appl., 191 (2022), 116399. https://doi.org/10.1016/j.eswa.2021.116399 doi: 10.1016/j.eswa.2021.116399
[45]	H. Huang, L. Lin, R. Tong, H. Hu, Q. Zhang, Y. Iwamoto, et al., UNet 3+: A full-scale connected UNet for medical image segmentation, in 2020 IEEE International Conference on Acoustics, Speech and Signal Processing, (2020). https://doi.org/10.48550/arXiv.2004.08790
[46]	M. Z. Alom, C. Yakopcic, T. M. Taha, V. K. Asari, Nuclei segmentation with recurrent residual convolutional neural networks based U-net(R2U-net), 2018-IEEE National Aerospace and Electronics Conference, (2018). https://doi.org/10.1109/NAECON.2018.8556686 doi: 10.1109/NAECON.2018.8556686
[47]	T. Shen, X. G. Li, Automatic polyp image segmentation and cancer prediction based on deep learning, Front. Oncol., 12 (2022), 1087438. https://doi.org/10.3389/fonc.2022.1087438 doi: 10.3389/fonc.2022.1087438
[48]	Z. Han, M. Jian, G. G. Wang, ConvUNeXt: An efficient convolution neural network for medical image segmentation, Knowl. Based Syst., 253 (2022), 109512. https://doi.org/10.1016/j.knosys.2022.109512 doi: 10.1016/j.knosys.2022.109512
[49]	R. Gu, G. Wang, T. Song, R. Huang, M. Aertsen, J. Deprest, et al., CA-Net: Comprehensive attention convolutional neural networks for explainable medical image segmentation, IEEE Trans. Med. Imaging, 40 (2020), 699–711. https://doi.org/10.48550/arXiv.2009.10549 doi: 10.48550/arXiv.2009.10549
[50]	J. Zhang, X. Lv, H. Zhang, B. Liu, AResU-Net: Attention residual U-Net for brain tumor segmentation, Symmetry, 12 (2020), 721. https://doi.org/10.3390/sym12050721 doi: 10.3390/sym12050721
[51]	X. Tong, J. Wei, B. Sun, S. Su, Z. Zuo, P. Wu, ASCU-Net: attention gate, spatial and channel attention u-net for skin lesion segmentation, Diagnostics, 11 (2021), 501. https://doi.org/10.3390/diagnostics11030501 doi: 10.3390/diagnostics11030501
[52]	J. Fu, J. Liu, H. J. Tian, Y. Li, Y. J. Bao, Z. W. Fang, et al., Dual attention network for scene segmentation, in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, (2019). https://doi.org/10.48550/arXiv.1809.02983
[53]	K. M. He, X. Y. Zhang, S. Q. Ren, J. Sun, Deep residual learning for image recognition, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), (2016). https://doi.org/10.1109/CVPR.2016.90 doi: 10.1109/CVPR.2016.90
[54]	M. Jun, J. N. Chen, M. Ng, R. Huang, Y. Li, C. Li, et al., Loss odyssey in medical image segmentation, Med. Image Anal., 71 (2021), 102035. https://doi.org/10.1016/j.media.2021.102035 doi: 10.1016/j.media.2021.102035
[55]	R. Wang, T. Lei, R. Cui, B. Zhang, H. Meng, A. K. Nandi, Medical image segmentation using deep learning: a survey, IET Image Process., 162 (2022), 1243–1267. https://doi.org/10.48550/arXiv.2009.13120 doi: 10.48550/arXiv.2009.13120
[56]	T. Y. Lin, P. Goyal, R. Girshick, K. He, P. Dollár, Focal loss for dense object detection, in Proceedings of the IEEE international conference on computer vision, (2017). https://doi.org/10.48550/arXiv.1708.02002
[57]	N. Codella, V. Rotemberg, P. Tschandl, M. E. Celebi, S. Dusza, D. Gutman, et al., Skin lesion analysis toward melanoma detection 2018: A challenge hosted by the international skin imaging collaboration (isic), preprint, arXiv: 1902.03368.
[58]	M. Yahyatabar, P. Jouvet, F. Cheriet, Dense-Unet: a light model for lung fields segmentation in Chest X-Ray images, in 42nd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), (2020). https://doi.org/10.1109/EMBC44109.2020.9176033
[59]	Y. Sun, F. K. Bi, Y. T. Gao, L. Chen, S. T. Feng, A Multi-Attention UNet for Semantic Segmentation in Remote Sensing Images, Symmetry, 14 (2022), 906. https://doi.org/10.3390/sym14050906 doi: 10.3390/sym14050906
[60]	Q. Jin, Z. Meng, T. D. Pham, Q. Chen, L. Wei, R. Su, Dunet: A deformable network for retinal vessel segmentation, Knowledge-Based Systems, 178 (2018), 149–162. https://doi.org/10.48550/arXiv.1811.01206 doi: 10.48550/arXiv.1811.01206
[61]	L. Li, M. Verma, Y. Nakashima, H. Nagahara, R. Kawasaki, Iternet: Retinal image segmentation utilizing structural redundancy in vessel networks, in The IEEE Winter Conference on Applications of Computer Vision, (2020). https://doi.org/10.1109/WACV45572.2020.9093621
[62]	S. A. Kamran, K. F. Hossain, A. Tavakkoli, S. L. Zuckerbrod, K. M. Sanders, S. A. Baker, RV-GAN: segmenting retinal vascular structure in fundus photographs using a novel multi-scale generative adversarial network, in International Conference on Medical Image Computing and Computer-Assisted Intervention, (2021). https://doi.org/10.48550/arXiv.2101.00535
[63]	Y. Zhang, H. Liu, Q. Hu, Transfuse: Fusing transformers and cnns for medical image segmentation, in International Conference on Medical Image Computing and Computer-Assisted Intervention, (2021). https://doi.org/10.48550/arXiv.2102.08005
[64]	J. Wei, Y. Hu, R. Zhang, Z. Li, S. K. Zhou, S. Cui, Shallow attention network for polyp segmentation, in International Conference on Medical Image Computing and Computer-Assisted Intervention, (2021). https://doi.org/10.48550/arXiv.2108.00882
[65]	J. M. J. Valanarasu, V. M. Patel, Unext: Mlp-based rapid medical image segmentation network, preprint, arXiv: 2203.04967.
[66]	R. Azad, M. Asadi-Aghbolaghi, M. Fathy, S. Escalera, Bi-directional ConvLSTM U-Net with densley connected convolutions, in Proceedings of the IEEE/CVF international conference on computer vision workshops, (2019). https://doi.org/10.48550/arXiv.1909.00166
[67]	X. G. Peng, D. L. Peng, MDA-Net: a medical image segmentation network combining dual-path attention mechanism, Small Microcomputer Syst., 43 (2022), 1–9. http://kns.cnki.net/kcms/detail/21.1106.tp.20220729.1534.034.html.

This article has been cited by:

Yakui Wu, Qiong Wu, Yue Zhang, Time decay estimates of solutions to a two-phase flow model in the whole space, 2024, 13, 2191-950X, 10.1515/anona-2024-0037

Reader Comments

Your name:*

Email:*
© 2023 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)