An Automatic Detection and Localization of Mammographic Microcalcifications ROI with Multi-Scale Features Using the Radiomics Analysis Approach.

Tariq Mahmood,Faheem Akhtar,Yan Pei,Azhar Imran,Muhammad Yaqub,Jianqiang Li

doi:10.3390/cancers13235916

Abstract

Simple SummaryBreast cancer is one of the foremost causes of cancer-related mortality in women. It is curable and controllable only if detected early. Microcalcifications in breast tissue are essential predictors for radiologists to detect early-stage breast cancer. This study proposes a method for detecting and classifying microcalcifications in mammogram images to predict breast lesions, using machine learning coupled with an interpretable radiomics approach. The method was evaluated using a publicly accessible dataset, which may aid radiologists and clinicians in identifying breast cancer in their regular clinical practices. This study contributes to the field of predictive modeling in healthcare.Microcalcifications in breast tissue can be an early sign of breast cancer, and play a crucial role in breast cancer screening. This study proposes a radiomics approach based on advanced machine learning algorithms for diagnosing pathological microcalcifications in mammogram images and provides radiologists with a valuable decision support system (in regard to diagnosing patients). An adaptive enhancement method based on the contourlet transform is proposed to enhance microcalcifications and effectively suppress background and noise. Textural and statistical features are extracted from each wavelet layer’s high-frequency coefficients to detect microcalcification regions. The top-hat morphological operator and wavelet transform segment microcalcifications, implying their exact locations. Finally, the proposed radiomic fusion algorithm is employed to classify the selected features into benign and malignant. The proposed model’s diagnostic performance was evaluated on the MIAS dataset and compared with traditional machine learning models, such as the support vector machine, K-nearest neighbor, and random forest, using different evaluation parameters. Our proposed approach outperformed existing models in diagnosing microcalcification by achieving an 0.90 area under the curve, 0.98 sensitivity, and 0.98 accuracy. The experimental findings concur with expert observations, indicating that the proposed approach is most effective and practical for early diagnosing breast microcalcifications, substantially improving the work efficiency of physicians.

Highlights

Breast cancer is the most common malignancy in women across the globe [1]
The results indicate that the microcalcification point coefficients are dispersed mainly on the second and third layer high-frequency wavelet decomposition coefficients selected for reconstruction, resulting in a noise-free breast image
The experimental findings reveal that the proposed model significantly outperformed with 0.98 sensitivity at a 1.2 false-positive per image (FPi) compared to other random forest (RF), K-nearest neighbor (K-NN), and support vector machine (SVM) studies

Summary

Introduction

Breast cancer is the most common malignancy in women across the globe [1]. Primarily mammograms, are the preferred diagnostic tools for early breast cancer screening, and have been proven to decrease the mortality rate by up to 30% [2]. Mammography detects about up to 80–90% of breast cancer patients without symptoms at an initial stage [3]. The Breast Imaging Reporting and Data System (BI-RADS) lexicon characterizes mammogram features by breast density, mass, calcification, asymmetry, lesion location, and related findings [4]. A mammogram distinguishes the breast density based on the recommended lexicon: fatty, scattered, heterogeneously dense, and highly dense. Suspicious calcifications in the breast are characterized as diffuse regional, linear, clustered, amorphous, and segmental [5].

Objectives

Methods

Results

Conclusion