11.deep Learning Applications in Medical Image Analysis-Brain Tumor
11.deep Learning Applications in Medical Image Analysis-Brain Tumor
11.deep Learning Applications in Medical Image Analysis-Brain Tumor
ABSTRACT:
INDEX TERMS Convolutional neural networks, medical image analysis, machine learning,
deep learning.
INTRODUCTION:
Machine learning algorithms have the potential to be invested deeply in all _elds of
medicine, from drug discovery to clinical decision making, signi_cantly altering the way
medicine is practiced. The success of machine learning algorithms at computer vision tasks in
recent years comes at an opportune time when medical records are increasingly digitalized. The
use of electronic health records (EHR) quadrupled from 11.8% to 39.6% amongst of_ce-based
physicians in the US from 2007 to 2012 [1]. Medical images are an integral part of a patient's
EHR and are currently analyzed by human radiologists, who are limited by speed, fatigue, and
experience. It takes years and great _nancial cost to train a quali_ed radiologist, and some health-
care systems outsource radiology reporting to lower-cost countries such as India via tele-
radiology. A delayed or erroneous diagnosis causes harm to the patient. Therefore, it is ideal for
medicalimage analysis to be carried out by an automated, accurate and ef_cient machine learning
algorithm.
EXISTING SYSTEM:
There is a myriad of imaging modalities, and the frequencyof their use is increasing.
Smith-Bindmanet al. [2] lookedat imaging use from 1996 to 2010 across six large
integratedhealthcare systems in the United States, involving30.9 million imaging examinations.
The authors found thatover the study period, CT, MRI and PET usage increased7.8%, 10% and
57% respectively.
The symbolic AI paradigm of the 1970s led to the developmentof rule-based, expert systems.
One early implementationin medicine was the MYCIN system by Shortliffe [3],which suggested
different regimes of antibiotic therapiesfor patients. Parallel to these developments, AI
algorithmsmoved from heuristics-based techniques to manual, handcraftedfeature extraction
techniques. and then to supervisedlearning techniques. Unsupervised machine learning methods
are also being researched, but the majority of the algorithmsfrom 2015-2017 in the published
literature have employedsupervised learning methods,
PROPOSED SYSTEM:
Currently, CNNs are the most researched machine learningalgorithms in medical image
analysis [4]. The reason for thisis that CNNs preserve spatial relationships when _lteringinput
images. As mentioned, spatial relationships are of crucialimportance in radiology, for example,
in how the edgeof a bone joins with muscle, or where normal lung tissueinterfaces with
cancerous tissue. As shown in Fig. 2., a CNNtakes an input image of raw pixels, and transforms
it viaConvolutional Layers, Recti_ed Linear Unit (RELU) Layersand Pooling Layers. This feeds
into a _nal Fully ConnectedLayer which assigns class scores or probabilities, thus classifyingthe
input into the class with the highest probability.
SOFTWARE REQUIREMENTS:
OS : Windows
Python IDE : python 2.7.x and above
Pycharm IDE,
Anaconda 3.5
Setup tools and pip to be installed for 3.6.x and above
HARDWARE REQUIREMENTS: