ABSTRACT
PURPOSE: The purpose of this study was to create a deep learning algorithm to infer the benign or malignant nature of breast nodules using two-dimensional B-mode ultrasound data initially marked as BI-RADS 3 and 4. MATERIALS AND METHODS: An ensemble of mask region-based convolutional neural networks (Mask-RCNN) combining nodule segmentation and classification were trained to explicitly localize the nodule and generate a probability of the nodule to be malignant on two-dimensional B-mode ultrasound. These probabilities were aggregated at test time to produce final results. Resulting inferences were assessed using area under the curve (AUC). RESULTS: A total of 460 ultrasound images of breast nodules classified as BI-RADS 3 or 4 were included. There were 295 benign and 165 malignant breast nodules used for training and validation, and another 137 breast nodules images used for testing. As a part of the challenge, the distribution of benign and malignant breast nodules in the test database remained unknown. The obtained AUC was 0.69 (95% CI: 0.57-0.82) on the training set and 0.67 on the test set. CONCLUSION: The proposed deep learning solution helps classify benign and malignant breast nodules based solely on two-dimensional ultrasound images initially marked as BIRADS 3 and 4.
Subject(s)
Algorithms , Neural Networks, Computer , Area Under Curve , Humans , UltrasonographyABSTRACT
We show here by means of scanning tunneling microscopy (STM) at the liquid/solid interface that paracetamol and benzocaine molecules bearing a long aliphatic chain can be immobilized on highly oriented pyrolitic graphite (HOPG) as perfectly ordered two-dimensional domains extending over several hundreds of nanometers. In both cases, high-resolution STM images reveal that compounds 1 and 2 self-assemble into parallel lamellae having a head-to-head arrangement. The paracetamol heads of 1 are in a zigzag position with entangled n-dodecyloxy side chains while benzocaine heads of compound 2 are perfectly aligned as a double row and have their palmitic side chains on either sides of the head alignment. We attribute the very long-range ordering of these two pro-drug derivatives on HOPG to the combined effects of intermolecular H-bonding on one side and Van der Waals interactions between aliphatic side chains and graphite on the other side. The 2D immobilization of pro-drug derivatives via a non-destructive physisorption mechanism could prove to be useful for applications such as drug delivery if it can be realized on a biocompatible substrate.