A multi stage deep learning model for accurate segmentation and classification of breast lesions in mammography

Sep 23, 2026

Uploaded by Fans

Mammography relies on radiologists to detect breast lesions, yet reliable detection and segmentation remain difficult because image quality and lesion characteristics vary. This work proposes a Multi-Stage Deep Lesion model (MSDLM) combining U-Net for segmentation, EfficientNetV2 B0 for deep feature extraction, and a domain CNN classifier for classification. The feature-cascade design targets computational efficiency and preserves the most relevant information. Validation on CBIS-DDSM and the Wisconsin breast cancer dataset shows 97.6% accuracy, 91.25% sensitivity, 95.75% AUC, and 85.59% IoU, outperforming baseline models.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages23

Reading Time58 min

AccessFree to download

Method Overview

Three-Unit MSDLM Architecture

Feature Cascade Design

Experimental Validation

Datasets

Metrics and Results

Context of Breast Cancer Detection

[www. nature.com/scientificreports](www. nature.com/scientificreports) OPEN A multi stage deep learning model for accurate segmentation and classification of breast lesions in mammography Sultan Ahmad1,2􀀍, Eali Stephen Neal Joshua3, Nakka Thirupathi Rao4, Rania M. Ghoniem5, Belayneh MatebieTaye7􀀍 & Salil Bharany6 Mammography is a routine imaging technique used by radiologists to detect breast lesions, such as tumors and lumps. Precise lesion detection is critical for early treatment and diagnosis planning. Lesion detection and segmentation are still problematic due to inconsistencies in image quality and lesion properties. Hence, this work presents a new Multi-Stage Deep Lesion model (MSDLM) for enhancing the efficacy of breast lesion segmentation and classification. The suggested model is a Three-Unit Two-Parameter Gaussian model with U-Net, EfficientNetV2 B0, and a domain CNN classifier. U-Net is utilized for lesion segmentation, EfficientNetV2 B0 is employed for image deep feature extraction, and a CNN classifier is used for lesion classification. The MSDLM feature cascade is designed to enhance computational efficiency while retaining the most relevant features for breast cancer detection and identifying the minimum number of features most important in the detection and classification of breast lesions in mammograms. The Multi-Stage Deep Learning Model (MSDLM) was validated using two benchmark datasets, CBIS-DDSM and the Wisconsin breast cancer dataset. Segmentation and classification tasks were indicated by accuracy values of 97.6%, indicating reliability in breast lesion detection. Its sensitivity of 91.25% indicates its reliability in detecting positive cases, a basic requirement in medical diagnosis. It also indicated an Area Under the Curve (AUC) of 95.75%, indicating overall diagnostic performance irrespective of thresholds. The Intersection over Union (IoU) of 85.59% verifies its reliability to detect lesion areas in mammograms. MSDLM with a Gaussian distribution provides precise localization and classification of breast lesions. The MSDLM model allows for improved information flow and feature refinement. The algorithm outperforms baseline models in both effectiveness and computational cost. Its performance on two diverse datasets verifies its generalizability and enables radiologists to receive exact automated diagnoses. Keywords Mammography, GAU-Net, Gaussian distribution, Transfer learning, U-Net, EfficientNetV2 B0, Breast mass segmentation, Accuracy, Segmentation Cancer is the primary cause of mortality worldwide, exceeding all other illnesses. The latest official data shows that breast cancer is now more common than lung cancer. Breast cancer is the primary cause of cancer mortality among females. By 20261, breast, bowel, and lung cancers will be the most frequently diagnosed kinds of cancer in American women. Since peaking in 19892, the incidence of breast cancer deaths in women has decreased by 43%. The annual decrease in breast cancer mortality fell by approximately 2–3% in the 1990 s to 1% in 20243. Non-invasive breast cancer may remain localized inside a particular organ or structure, such as a duct or 1Department of Computer Science, College of Computer Engineering and Sciences, Prince Sattam Bin Abdulaziz University, Alkharj 11942, Saudi Arabia. 2School of Computer Science and Engineering, Lovely Professional University, Phagwara 144411, Punjab, India. 3Department of Computer Science & Engineering, Gandhi Institute of Technology and Management (GITAM) (Deemed to be University), Visakhapatnam, Andhra Pradesh, India. 4Department of Computer Science & Engineering, Vignan’s Institute of Information Technology (A), Visakhapatnam 530049, Andhra Pradesh, India. 5Department of Information Technology, College of Computer and Information Sciences, Princess Nourah bint Abdulrahman University, P.O. Box 84428, Riyadh 11671, Saudi Arabia. 6Chitkara University Institute of Engineering and Technology,

FAQs

What problem does the Multi-Stage Deep Lesion model address in mammography?

It targets the challenge of inconsistent image quality and varying lesion properties that make lesion detection and segmentation difficult, aiming to improve both segmentation and classification reliability.

How does MSDLM perform lesion segmentation and classification?

U-Net performs lesion segmentation, EfficientNetV2 B0 extracts deep features, and a domain CNN classifier outputs lesion classification using the cascaded multi-stage feature flow.

Which datasets were used to validate the model and what key performance metrics were reported?

Validation used CBIS-DDSM and the Wisconsin breast cancer dataset, reporting accuracy of 97.6%, sensitivity of 91.25%, AUC of 95.75%, and IoU of 85.59%.

Sep 23, 2026

Uploaded by Fans