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

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.

A Systematic Literature Review on Mammography: Deep Learning in Redefining Breast Cancer Diagnosis for the Asian Perspective

“…This technique improves the information about breast mass. (Ull et al, 2023;Bandaru et al, 2022a;Duong et al, 2023;Kahnouei et al, 2022;P and V, 2022;Li et al, 2019aLi et al, , 2021cMahmood et al, 2022;Wang et al, 2022;Yao et al, 2022), (Tsochatzidis et al, 2021;Ahmed et al, 2020;Fathy and Ghoneim, 2019;Frazer et al, 2021;Sannasi Chakravarthy and Rajaguru, 2021;Chakravarthy and Rajaguru, 2022;Al-Antari et al, 2020;Al-Mansour et al, 2022;Alruwaili and Gouda, 2022;Altameem et al, 2022;Altaf, 2021;Cao et al, 2020;Al-Tam et al, 2022;Bandaru et al, 2022b;López-Cabrera et al, 2020;Saber et al, 2021;Chougrad et al, 2018;Falconi et al, 2020;Jafarzadeh Ghoushchi et al, 2021;Hanis et al, 2023;Lin et al, 2022;Mokni and Haoues, 2022;Mudeng et al, 2022;Khan and Masala, 2023;Oza et al, 2023;Prodan et al, 2023;Ragab et al, 2021;Yu et al, 2023b;Gerbasi et al, 2023;Shanker and Vadivel, 2022;Zhang and Wang, 2019;Adedigba et al, 2022;…”

“…Graph CNN (GCN) (Duong et al, 2023;Yao et al, 2022), (Zhang et al, 2021b) These networks are designed for dealing with graph-structured data, with entities represented as nodes and relationships as edges. GCNs efficiently detect dependencies among interconnected nodes.…”

“…This method classifies pixels in an image into various regions or objects by considering the intensity values compared to a selected threshold. Morphological operations + Otsu + Color Space (Yao et al, 2022), (Cai et al, 2021;Chugh et al, 2022;Hekal et al, 2021;Singh et al, 2023) Otsu's method efficiently calculates an optimal threshold for separating foreground and background. Also, multilevel thresholding using otsu's is possible (Singh et al, 2023).…”