Delving into Ipsilateral Mammogram Assessment under Multi-View Network

TL;DR

This paper investigates multi-view fusion strategies in ipsilateral mammogram analysis, demonstrating that middle-layer fusion improves model generalization and performance across datasets.

Contribution

It introduces and evaluates diverse fusion strategies within a multi-view network, highlighting the effectiveness of middle-layer fusion for mammogram assessment.

Findings

Middle-layer fusion yields the best balance and performance.

Fusion strategies significantly impact model generalization.

Average and concatenate fusion improve macro F1-Score on datasets.

Abstract

In many recent years, multi-view mammogram analysis has been focused widely on AI-based cancer assessment. In this work, we aim to explore diverse fusion strategies (average and concatenate) and examine the model's learning behavior with varying individuals and fusion pathways, involving Coarse Layer and Fine Layer. The Ipsilateral Multi-View Network, comprising five fusion types (Pre, Early, Middle, Last, and Post Fusion) in ResNet-18, is employed. Notably, the Middle Fusion emerges as the most balanced and effective approach, enhancing deep-learning models' generalization performance by +2.06% (concatenate) and +5.29% (average) in VinDr-Mammo dataset and +2.03% (concatenate) and +3% (average) in CMMD dataset on macro F1-Score. The paper emphasizes the crucial role of layer assignment in multi-view network extraction with various strategies.