A Novel Lightweight Deep Learning-Based Histopathological Image Classification Model for IoMT
- 8 June 2021
- journal article
- research article
- Published by Springer Nature in Neural Processing Letters
- Vol. 55 (1) , 205-228
Abstract
The unavailability of appropriate mechanisms for timely detection of diseases and successive treatment causes the death of a large number of people around the globe. The timely diagnosis of grave diseases like different forms of cancer and other life-threatening diseases can save a valuable life or at least extend the life span of an afflicted individual. The advancement of the Internet of Medical Things (IoMT) enabled healthcare technologies can provide effective medical facilities to the population and contribute greatly towards the recuperation of patients. The usage of IoMT in the diagnosis and study of histopathological images can enable real-time identification of diseases and corresponding remedial actions can be taken to save an affected individual. This can be achieved by the use of imaging apparatus with the capacity of auto-analysis of captured images. However, most deep learning-based image classifying models are bulk in size and are inappropriate for use in IoT based imaging devices. The objective of this research work is to design a deep learning-based lightweight model suitable for histopathological image analysis with appreciable accuracy. This paper presents a novel lightweight deep learning-based model "ReducedFireNet", for auto-classification of histopathological images. The proposed method attained a mean accuracy of 96.88% and an F1 score of 0.968 on evaluating an actual histopathological image data set. The results are encouraging, considering the complexity of histopathological images. In addition to the high accuracy the lightweight design (size in few KBs) of the ReducedFireNet model, makes it suitable for IoMT imaging equipment. The simulation results show the proposed model has computational requirement of 0.201 GFLOPS and has a mere size of only 0.391 MB.Keywords
This publication has 33 references indexed in Scilit:
- Extensible Disease Diagnosis Cloud Platform with Medical Sensors and IoT Devices2015
- Health Monitoring and Management Using Internet-of-Things (IoT) Sensing with Cloud-Based Processing: Opportunities and Challenges2015
- Classification of lung image and nodule detection using fuzzy inference system2015
- Classification of Histological Images Based on the Stationary Wavelet TransformJournal of Physics: Conference Series, 2015
- Marker-Controlled Watershed for Lesion Segmentation in MammogramsJournal of Digital Imaging, 2011
- Automatic Classification of Lymphoma Images With Transform-Based Global FeaturesIEEE Transactions on Information Technology in Biomedicine, 2010
- Unsupervised Learning of Shape Manifolds2007
- Ipsilateral-mammogram computer-aided detection of breast cancerComputerized Medical Imaging and Graphics, 2004
- Multiwavelet grading of pathological images of prostateIEEE Transactions on Biomedical Engineering, 2003
- Computer‐aided diagnosis: Automatic detection of malignant masses in digitized mammogramsMedical Physics, 1998