Lightweight Deep Learning Pipeline for Detection, Segmentation and Classification of Breast Cancer Anomalies

• Type: Article in International Conference Proceedings Book
• Date: 2019

Authors

Oliveira, HS (Author) Other; Teixeira, JF (Author) Other; Oliveira, HP (Author) FCUP

Conference proceedings International

Title: IMAGE ANALYSIS AND PROCESSING - ICIAP 2019, PT II

Pages: 707-715

20th International Conference on Image Analysis and Processing (ICIAP)

Univ Trento, Fac Law, Trento, ITALY, SEP 09-13, 2019

Indexing

ISI Web of Knowledge - 6 Citations

Scopus - 8 Citations

Other information

Authenticus ID: P-00R-4FX

DOI: 10.1007/978-3-030-30645-8_64

Abstract (EN): The small amount of public available medical images hinders the use of deep learning techniques for mammogram automatic diagnosis. Deep learning methods require large annotated training sets to be effective, however medical datasets are costly to obtain and suffer from large variability. In this work, a lightweight deep learning pipeline to detect, segment and classify anomalies in mammogram images is presented. First, data augmentation using the ground-truth annotation is performed and used by a cascade segmentation and classification methods. Results are obtained using the INbreast public database in the context of lesion detection and BI-RADS classification. Moreover, a pre-trained Convolutional Neural Network using ResNet50 is modified to generate the lesion regions proposals followed by a false positive reduction and contour refinement stages while a pre-trained VGG16 network is fine-tuned to classify mammograms. The detection and segmentation stage results show that the cascade configuration achieves a DICE of 0.83 without massive training while the multi-class classification exhibits an MAE of 0.58 with data augmentation.

Language: English

Type (Professor's evaluation): Scientific

No. of pages: 9