Saved in:
Bibliographic Details
Main Authors: Ben-Artzi, Gil, Daragma, Feras, Mahpod, Shahar
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2411.10894
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929594744963072
author Ben-Artzi, Gil
Daragma, Feras
Mahpod, Shahar
author_facet Ben-Artzi, Gil
Daragma, Feras
Mahpod, Shahar
contents While state-of-the-art models for breast cancer detection leverage multi-view mammograms for enhanced diagnostic accuracy, they often focus solely on visual mammography data. However, radiologists document valuable lesion descriptors that contain additional information that can enhance mammography-based breast cancer screening. A key question is whether deep learning models can benefit from these expert-derived features. To address this question, we introduce a novel multi-modal approach that combines textual BI-RADS lesion descriptors with visual mammogram content. Our method employs iterative attention layers to effectively fuse these different modalities, significantly improving classification performance over image-only models. Experiments on the CBIS-DDSM dataset demonstrate substantial improvements across all metrics, demonstrating the contribution of handcrafted features to end-to-end.
format Preprint
id arxiv_https___arxiv_org_abs_2411_10894
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Deep BI-RADS Network for Improved Cancer Detection from Mammograms
Ben-Artzi, Gil
Daragma, Feras
Mahpod, Shahar
Computer Vision and Pattern Recognition
While state-of-the-art models for breast cancer detection leverage multi-view mammograms for enhanced diagnostic accuracy, they often focus solely on visual mammography data. However, radiologists document valuable lesion descriptors that contain additional information that can enhance mammography-based breast cancer screening. A key question is whether deep learning models can benefit from these expert-derived features. To address this question, we introduce a novel multi-modal approach that combines textual BI-RADS lesion descriptors with visual mammogram content. Our method employs iterative attention layers to effectively fuse these different modalities, significantly improving classification performance over image-only models. Experiments on the CBIS-DDSM dataset demonstrate substantial improvements across all metrics, demonstrating the contribution of handcrafted features to end-to-end.
title Deep BI-RADS Network for Improved Cancer Detection from Mammograms
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2411.10894