A depth-wise separable residual neural network for <em>PCDH8</em> status prediction in thyroid cancer pathological images

Linlin Qi; Xiangyu Li; Zhihong Liu; Pei Zhang; Liangliang Liu

doi:10.1016/j.intonc.2025.09.003

您当前的位置：

首页 >

文章列表页 >

A depth-wise separable residual neural network for PCDH8 status prediction in thyroid cancer pathological images

更新时间：2026-06-03

- A depth-wise separable residual neural network for PCDH8 status prediction in thyroid cancer pathological images
- Intelligent Oncology Issue 4, Pages: 290-298(2025)
- 作者机构：
  
  College of Information and Management Science, Henan Agricultural University,Henan,Zhengzhou,China,450046
- 作者简介：
- 基金信息：
- DOI：10.1016/j.intonc.2025.09.003
  CLC：
- Received：21 June 2025，
  
  Online First：26 September 2025，
  
  Published：15 October 2025
- 稿件说明：
移动端阅览
Linlin Qi, Xiangyu Li, Zhihong Liu, et al. A depth-wise separable residual neural network for PCDH8 status prediction in thyroid cancer pathological images[J]. Intelligent Oncology, 2025, (4): 290-298. DOI: 10.1016/j.intonc.2025.09.003.
DOI：

Linlin Qi, Xiangyu Li, Zhihong Liu, et al. A depth-wise separable residual neural network for PCDH8 status prediction in thyroid cancer pathological images[J]. Intelligent Oncology, 2025, (4): 290-298. DOI: 10.1016/j.intonc.2025.09.003. DOI：

摘要

IntroductionAccurate prediction of protocadherin 8 (

PCDH8

) gene expression status from whole-slide images (WSIs) is critical for thyroid cancer diagnosis and prognosis

PCDH8

overexpression is associated with tumor aggressiveness and poor outcomes. Existing methods for

PCDH8

detection are often costly

time-consuming

or require specialized expertise. To address these limitations

we developed a novel depth-wise separable residual neural network (DSRNet) for noninvasive

PCDH8

status prediction directly from WSIs.Materials and methodsWe collected 403 thyroid cancer WSIs from The Cancer Genome Atlas (TCGA)

with

PCDH8

expression status classified as high or low based on median expression values. Each WSI was divided into 512 × 512 pixel tiles

with the top 100 non-white tiles selected per slide. DSRNet integrates depth-wise separable convolutions

residual connections

and a deformable convolutional pyramid pooling module to efficiently capture multiscale and long-range features in g

igapixel WSIs. The model was trained using tenfold cross-validation.ResultsDSRNet achieved state-of-the-art performance with 92.76% accuracy

91.92% precision

92.69% recall

and 0.93 area under the curve on the thyroid cancer dataset (TCGA-THCA)

significantly outperforming leading convolutional neural networks and Transformer models. Ablation studies confirmed the contributions of each component

and attention visualization showed that DSRNet focuses on biologically relevant regions. The model also generalized well to a breast cancer dataset (TCGA-BRCA)

achieving 89.13% accuracy.ConclusionsWe developed DSRNet

a deep learning-based model for predicting

PCDH8

status directly from routine hematoxylin and eosin-stained pathological images. DSRNet combines the efficiency of convolutional operations with enhanced long-range dependency modeling

providing a noninvasive

accurate

and interpretable tool for auxiliary thyroid cancer diagnosis and prognosis. The results demonstrate its strong potential for clinical translation

though further multicenter validation is warranted.

Abstract

IntroductionAccurate prediction of protocadherin 8 (

PCDH8

) gene expression status from whole-slide images (WSIs) is critical for thyroid cancer diagnosis and prognosis

PCDH8

overexpression is associated with tumor aggressiveness and poor outcomes. Existing methods for

PCDH8

detection are often costly

time-consuming

or require specialized expertise. To address these limitations

we developed a novel depth-wise separable residual neural network (DSRNet) for noninvasive

PCDH8

status prediction directly from WSIs.Materials and methodsWe collected 403 thyroid cancer WSIs from The Cancer Genome Atlas (TCGA)

with

PCDH8

expression status classified as high or low based on median expression values. Each WSI was divided into 512 × 512 pixel tiles

with the top 100 non-white tiles selected per slide. DSRNet integrate

s depth-wise separable convolutions

residual connections

and a deformable convolutional pyramid pooling module to efficiently capture multiscale and long-range features in gigapixel WSIs. The model was trained using tenfold cross-validation.ResultsDSRNet achieved state-of-the-art performance with 92.76% accuracy

91.92% precision

92.69% recall

and 0.93 area under the curve on the thyroid cancer dataset (TCGA-THCA)

significantly outperforming leading convolutional neural networks and Transformer models. Ablation studies confirmed the contributions of each component

and attention visualization showed that DSRNet focuses on biologically relevant regions. The model also generalized well to a breast cancer dataset (TCGA-BRCA)

achieving 89.13% accuracy.ConclusionsWe developed DSRNet

a deep learning-based model for predicting

PCDH8

status directly from routine hematoxylin and eosin-stained pathological images. DSRNet combines the efficiency of convolutional operations with enhanced long-range dependency modeling

providing a noninvasive

accurate

and interpretable tool for auxiliary thyroid cancer diagnosis and prognosis. The results demonstrate its strong potential for clinical translation

though further multicenter validation is warranted.

关键词

Keywords

references

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Computational pathology: A comprehensive review of recent developments in digital and intelligent pathology

AWCDL: Automatic weight calibration deep learning for detecting HER2 status in whole-slide breast cancer image

TSMIL: Transformer-based structured low-rank end-to-end multi-instance learning network for renal cell carcinoma classification in whole-slide images

A multiscale residual dense fusion network for nuclear medical image fusion

From morphology to function: Advances in multimodal imaging for pulmonary function prediction

Related Author

Qinqin Huang

Shen’ao Wu

Zhenkai Ou

Yi Gao

Shuai Wang

Xiaojing Guo

Jinlu Ma

Ning Wu

Related Institution

School of Biomedical Engineering, Shenzhen University Medical School, Shenzhen University

Guangdong Provincial Key Laboratory of Mathematical and Neural Dynamical Systems

Marshall Laboratory of Biomedical Engineering

Department of Radiation Oncology, Shandong Cancer Hospital and Institute, Shandong First Medical University and Shandong Academy of Medical Sciences

Department of Biochemistry and Molecular Biology, Key Laboratory of Breast Cancer Prevention and Therapy, Tianjin Clinical Research Center for Cancer, Ministry of Education, National Clinical Research Center for Cancer, Tianjin Medical University Cancer Institute & Hospital

⁰

A depth-wise separable residual neural network for PCDH8 status prediction in thyroid cancer pathological images

DOI：10.1016/j.intonc.2025.09.003

摘要

Abstract

关键词

Keywords

references