Novel hybrid approach combining ANN and MRA for PET volume segmentation

Abstract

Medical volume segmentation is an essential stage in volume processing. This stage is important for tumour classification and quantification in medical volumes particularly in positron emission tomography (PET) imaging. Analysing PET volumes at early stage of illness is important for radiotherapy planning, tumour diagnosis, and fast recovery. There are many techniques for segmenting medical volumes, in which some of the approaches have poor accuracy and require a lot of time for analysing large medical volumes. In this paper, a novel hybrid approach (HA) combining artificial neural network (ANN) with multiresolution analysis (MRA) for segmenting oncological PET data aiming at providing an accurate quantitative analysis tool is proposed. Proposing artificial intelligence (AI) technologies can provide better accuracy and save decent amount of time. The proposed approach has been evaluated against other medical volume segmentation techniques such as thresholding, clustering, and multiscale Markov random field model. The proposed approach has shown promising results in terms of the detection and quantification of the region of interest (ROI) and tumour, in phantom and clinical PET volumes respectively.