Segmentation of infrared images: A new technology for early detection of breast diseases

Abstract

In a first stage a cancer promotes an intense process of vascularization at the affected area increasing blood flow and modifying the local temperature of the body. Using a thermal camera, the infrared radiation emitted by the human body can be captured and then used in the measuring of body temperature, turning the results into an image. Moreover, thermography can detect suspicious regions in patients of any age, even in cases of dense breasts, where the detection of an abnormality cannot be accomplished by others exams. A fundamental step in the use of thermal images is the development of computer aided diagnosis (CAD) systems. These could allow the execution of exams by technicians, following well established routines and protocols, as already occurs in mammography exams, allowing doctors to have a greater possibility of dedication in the analysis and in the meaning of the exams. In this work, an automatic detection of the regions of interest (ROI) is proposed and compared with segmentations performed manually. This work presents a methodology for the automatic segmentation of lateral breast thermal images. For the evaluation of the results, different groups of ground truth are generated, which are available on the internet, in order to allow the verification of the results' correctness. Finally, the obtained results by the proposed methodology for the 328 images used in this work are demonstrated. The results showed average values of accuracy

Medical diagnostic support system based on breast thermography using Raspberry Pi and cloud computing

“…One of the major problems with the use of thermography, especially when using a computer aided diagnostic system, is the precise segmentation of the region of interest (right and left breast) from the 789 background (extra mammary organs). Several segmentation techniques have been proposed in the literature to solve this problem, but they obtained moderate success rates [6], [15]. In an attempt to properly segment the breast regions, we used a deep learning U-net algorithm [16] which has showed great performance in similar pattern recognition and computer vision tasks [11], [17]- [19].…”

A Systematic Review of Breast Cancer Detection Using Thermography and Neural Networks

“…In [36], automatic detection of the regions of interest is proposed and compared with segmentations performed manually. The work presented a methodology for the automatic segmentation of lateral breast thermal images.…”

Wearable infra-red pre-screening smartbra for early detection of breast cancer

“…The reproducibility of the thermal pattern is important if medical infrared thermography is to be used as a screening tool. Thus, a three (3) days pre-study was conducted to evaluate the day-to-day repeatability of the device [37]. This was necessary as reliable measurements have a substantial impact on the diagnosis and interpretation of pathophysiological abnormalities [31].…”