End-to-End Calcification Distribution Pattern Recognition for Mammograms: An Interpretable Approach with GNN

Melissa Min-Szu Yao; Hao Du; Mikael Hartman; Wing P Chan; Mengling Feng

doi:10.3390/diagnostics12061376

End-to-End Calcification Distribution Pattern Recognition for Mammograms: An Interpretable Approach with GNN

Diagnostics (Basel). 2022 Jun 2;12(6):1376. doi: 10.3390/diagnostics12061376.

Authors

Melissa Min-Szu Yao^{1

2}, Hao Du^{3

4}, Mikael Hartman^{3

4

5}, Wing P Chan^{1

2

6}, Mengling Feng^{1

4

7}

Affiliations

¹ Department of Radiology, Wan Fang Hospital, Taipei Medical University, Taipei 116, Taiwan.
² Department of Radiology, School of Medicine, College of Medicine, Taipei Medical University, Taipei 110, Taiwan.
³ Saw Swee Hock School of Public Health, National University of Singapore, Singapore 117549, Singapore.
⁴ National University Health System, Singapore 119228, Singapore.
⁵ Department of Surgery, Yong Loo Lin School of Medicine, National University of Singapore, Singapore 117549, Singapore.
⁶ Medical Innovation Development Center, Wan Fang Hospital, Taipei Medical University, Taipei 116, Taiwan.
⁷ Institute of Data Science, National University of Singapore, Singapore 117602, Singapore.

Abstract

Purpose: We aimed to develop a novel interpretable artificial intelligence (AI) model algorithm focusing on automatic detection and classification of various patterns of calcification distribution in mammographic images using a unique graph convolution approach. Materials and methods: Images from 292 patients, which showed calcifications according to the mammographic reports and diagnosed breast cancers, were collected. The calcification distributions were classified as diffuse, segmental, regional, grouped, or linear. Excluded were mammograms with (1) breast cancer with multiple lexicons such as mass, asymmetry, or architectural distortion without calcifications; (2) hidden calcifications that were difficult to mark; or (3) incomplete medical records. Results: A graph-convolutional-network-based model was developed. A total of 581 mammographic images from 292 cases of breast cancer were divided based on the calcification distribution pattern: diffuse (n = 67), regional (n = 115), group (n = 337), linear (n = 8), or segmental (n = 54). The classification performances were measured using metrics including precision, recall, F1 score, accuracy, and multi-class area under the receiver operating characteristic curve. The proposed model achieved a precision of 0.522 ± 0.028, sensitivity of 0.643 ± 0.017, specificity of 0.847 ± 0.009, F1 score of 0.559 ± 0.018, accuracy of 64.325 ± 1.694%, and area under the curve of 0.745 ± 0.030; thus, the method was found to be superior compared to all baseline models. The predicted linear and diffuse classifications were highly similar to the ground truth, and the predicted grouped and regional classifications were also superior compared to baseline models. The prediction results are interpretable using visualization methods to highlight the important calcification nodes in graphs. Conclusions: The proposed deep neural network framework is an AI solution that automatically detects and classifies calcification distribution patterns on mammographic images highly suspected of showing breast cancers. Further study of the AI model in an actual clinical setting and additional data collection will improve its performance.

Keywords: artificial intelligence; calcifications; deep learning; graph convolution network; mammography.

Grants and funding

This research was funded by Ministry of Education under the Higher Education Sprout Project in Taiwan, grant numbers DP2-109-21121-01-A-03, DP2-110-21121-01-A-03, and DP2-111-21121-01-A-11, the National Research Foundation Singapore under its AI Singapore program, grant number AISG-GC-2019-001-2A, and an NMRC Health Service Research Grant, grant number MOH-000030-00.