A machine learning approach on multiscale texture analysis for breast microcalcification diagnosis

Abstract

Background: Screening programs use mammography as primary diagnostic tool for detecting breast cancer at an early stage. The diagnosis of some lesions, such as microcalcifications, is still difficult today for radiologists. In this paper, we proposed an automatic binary model for discriminating tissue in digital mammograms, as support tool for the radiologists. In particular, we compared the contribution of different methods on the feature selection process in terms of the learning performances and selected features. Results: For each ROI, we extracted textural features on Haar wavelet decompositions and also interest points and corners detected by using Speeded Up Robust Feature (SURF) and Minimum Eigenvalue Algorithm (MinEigenAlg). Then a Random Forest binary classifier is trained on a subset of a sub-set features selected by two different kinds of feature selection techniques, such as filter and embedded methods. We tested the proposed model on 260 ROIs extracted from digital mammograms of the BCDR public database. The best prediction performance for the normal/abnormal and benign/malignant problems reaches a median AUC value of 98.16% and 92.08%, and an accuracy of 97.31% and 88.46%, respectively. The experimental result was comparable with related work performance. Conclusions: The best performing result obtained with embedded method is more parsimonious than the filter one. The SURF and MinEigen algorithms provide a strong informative content useful for the characterization of microcalcification clusters.

Feasibility, Image Quality and Clinical Evaluation of Contrast-Enhanced Breast MRI Performed in a Supine Position Compared to the Standard Prone Position

“…Although these differences can be minimized through the use of computer-aided detection systems in MRI, as documented in the literature for other breast imaging methods [24][25][26][27][28][29], our study demonstrated that there is a significant difference in all the breast lesions in the images from the prone and the supine position with the exception of the fibroadenomas, the single papilloma, the lymph nodes and a silicoma, benign formations of the breast and invasive ductal carcinoma. In the remaining cases, a significant variation in the extension was observed, which generally consisted of a reduction in the extension in the prone compared to the supine position.…”

Detection and Diagnosis of Breast Cancer Using Artificial Intelligence Based Assessment of Maximum Intensity Projection Dynamic Contrast-Enhanced Magnetic Resonance Images

“…The performance of the model depends on how the data are split, which could lead to errors in its evaluation. An N-fold cross-validation method is potentially more effective for deep learning based on multiple training-validation splits and should provide a better indication of how well a model would perform given new previously unseen data [27,28]. An N-fold approach is currently beyond the capability of our resources, but future research will investigate whether an n-fold cross-validation method can be used to construct a highly accurate and stable AI system.…”

Elite VABB 13G: A New Ultrasound-Guided Wireless Biopsy System for Breast Lesions. Technical Characteristics and Comparison with Respect to Traditional Core-Biopsy 14–16G Systems

“…The microcalcification clusters, which are indicated in the VABB biopsy, if visible at ultrasound, are also easily and immediately sampled with Elite VABB, thus reducing the percentage of the stereotaxic VABBs and promoting a saving in terms of cost, time, and personnel (due to the absence of a radiological technician). This acknowledgement could increase further in the future in the event of development in automatic systems of detection and characterization of such specimens even in ultrasound as present in mammography [38][39][40][41][42].…”