‭PRIMITIVE DIAGNOSIS OF‬

‭RESPIRATORY DISEASES‬

‭Team Members‬

‭ .Bren Joel -CSE‬

S
‭S.Benazeer -CSE‬
‭N.Sasank -CSE‬
‭P.Gnanesh -CSE‬
 ‭ABSTRACT‬

‭ owadays‬‭for‬‭identifying‬‭or‬‭predicting‬‭any‬‭diseases‬‭on‬‭human‬‭beings,‬‭we‬‭should‬‭have‬‭a‬‭proper‬
N
‭diagnosis‬ ‭for‬ ‭predicting‬ ‭the‬ ‭disease‬ ‭which‬ ‭is‬ ‭present‬ ‭in‬ ‭that‬ ‭human‬ ‭body.‬ ‭In‬ ‭general‬ ‭for‬
‭prediction‬ ‭of‬ ‭diseases‬ ‭we‬ ‭try‬ ‭to‬ ‭use‬ ‭either‬ ‭X-Ray,‬ ‭CT‬ ‭or‬ ‭MRI‬ ‭scan‬ ‭techniques‬ ‭for‬ ‭taking‬
‭decisions‬ ‭on‬ ‭that‬ ‭appropriate‬ ‭disease.‬ ‭In‬ ‭general‬ ‭medical‬ ‭people‬ ‭need‬ ‭complete‬ ‭knowledge‬ ‭on‬
‭that‬‭appropriate‬‭domain‬‭to‬‭find‬‭out‬‭the‬‭abnormality‬‭which‬‭is‬‭present‬‭in‬‭human‬‭beings.‬‭As‬‭we‬‭all‬
‭know‬ ‭that‬ ‭India‬ ‭tops‬ ‭the‬ ‭world‬ ‭for‬ ‭having‬ ‭more‬ ‭deaths‬ ‭due‬ ‭to‬ ‭lung‬ ‭diseases.‬‭After‬‭the‬‭second‬
‭highest‬‭cause‬‭of‬‭deaths‬‭in‬‭India‬‭due‬‭to‬‭heart‬‭disease,‬‭this‬‭lung‬‭disease‬‭is‬‭one‬‭which‬‭is‬‭increasing‬
‭its‬ ‭rank‬ ‭more‬ ‭and‬ ‭more.‬ ‭In‬ ‭order‬ ‭to‬ ‭reduce‬ ‭that‬ ‭problem‬‭early‬‭diagnosis‬‭and‬‭treatment‬‭of‬‭lung‬
‭diseases‬ ‭is‬ ‭critical‬ ‭to‬ ‭prevent‬ ‭complications‬ ‭including‬ ‭death.‬ ‭Normally‬ ‭for‬ ‭finding‬ ‭the‬
‭abnormality‬‭present‬‭in‬‭lungs,‬‭lung‬ ‭X-ray‬‭is‬‭playing‬‭a‬‭very‬‭important‬‭role‬‭to‬‭detect‬‭the‬‭complete‬
‭information‬ ‭about‬ ‭the‬ ‭lungs.‬ ‭In‬ ‭this‬ ‭project‬ ‭we‬ ‭try‬ ‭to‬ ‭present‬ ‭an‬ ‭effective‬ ‭way‬ ‭for‬ ‭expert‬
‭diagnosis‬ ‭of‬ ‭lung‬ ‭diseases‬ ‭using‬ ‭deep‬ ‭learning‬ ‭models.This‬ ‭will‬ ‭especially‬ ‭benefit‬ ‭rural‬ ‭areas‬
‭where‬ ‭continuous‬ ‭medication‬ ‭is‬ ‭not‬ ‭easily‬ ‭available.We‬ ‭conclude‬ ‭by‬ ‭discussing‬ ‭research‬
‭obstacles,‬ ‭emerging‬ ‭trends,‬ ‭and‬ ‭possible‬ ‭future‬ ‭directions‬ ‭for‬ ‭improving‬ ‭some‬ ‭more‬
‭advancement.‬
 ‭INTRODUCTION‬

‭1.1 GENERAL BACKGROUND‬

‭ he‬‭chest‬‭is‬‭the‬‭most‬‭important‬‭part‬‭of‬‭the‬‭body‬‭as‬‭it‬
T
‭contains‬‭the‬‭respiration‬‭organs,‬‭which‬‭are‬‭responsible‬‭for‬‭sustaining‬‭the‬‭important‬‭life‬‭functions‬
‭of‬ ‭the‬ ‭body.‬ ‭The‬ ‭count‬ ‭of‬ ‭people‬ ‭being‬ ‭diagnosed‬ ‭with‬ ‭a‬ ‭chest‬ ‭disease‬‭globally‬‭is‬‭in‬‭millions.‬
‭Chest‬ ‭X-rays‬ ‭are‬ ‭considered‬ ‭to‬ ‭be‬ ‭one‬ ‭of‬ ‭the‬ ‭most‬ ‭cost-effective‬ ‭medical‬ ‭image‬ ‭examinations‬
‭available.‬ ‭The‬ ‭one‬ ‭who‬ ‭specializes‬ ‭in‬ ‭diagnosing‬ ‭diseases‬ ‭using‬ ‭X-rays‬ ‭is‬ ‭Radiologists.‬ ‭To‬
‭discover‬‭possible‬‭diseases‬‭in‬‭the‬‭lungs‬‭the‬‭radiologists‬‭spend‬‭a‬‭lot‬‭of‬‭time‬‭diagnosing‬‭chest‬‭X-ray‬
‭images.‬ ‭The‬ ‭knowledge‬ ‭of‬ ‭anatomical‬ ‭principles,‬ ‭physiology‬ ‭and‬ ‭pathology,‬ ‭and‬ ‭also‬ ‭keen‬
‭analysis‬‭are‬‭needed‬‭for‬‭diagnosing‬‭X-rays.‬‭Developing‬‭a‬‭system‬‭for‬‭analyzing‬‭chest‬‭X-ray‬‭could‬
‭make‬‭a‬‭huge‬‭impact‬‭on‬‭the‬‭radiologists.‬‭A‬‭chest‬‭X-ray‬‭produces‬‭a‬‭black-and-white‬‭image‬‭of‬‭the‬
‭organs‬‭in‬‭the‬‭chest.‬‭Tissues‬‭like‬‭lungs‬‭which‬‭consist‬‭of‬‭air‬‭appear‬‭as‬‭black‬‭and‬‭bones‬‭like‬‭dense‬
‭tissues‬ ‭absorb‬ ‭X-rays‬ ‭that‬ ‭appear‬ ‭white‬ ‭in‬ ‭images.‬ ‭The‬ ‭gray‬ ‭appears‬ ‭as‬ ‭tissues.‬ ‭Some‬ ‭of‬ ‭the‬
‭diseases‬ ‭that‬ ‭are‬ ‭diagnosed‬ ‭using‬ ‭chest‬ ‭x-ray‬ ‭images‬ ‭are‬ ‭heart‬ ‭diseases,‬ ‭pneumothorax,‬
‭bronchitis,‬ ‭fractures,‬ ‭etc.‬ ‭However,‬ ‭reading‬ ‭X‬ ‭ray‬ ‭images‬ ‭can‬ ‭be‬ ‭tricky‬ ‭and‬ ‭requires‬ ‭domain‬
‭expertise‬ ‭and‬ ‭experience.An‬ ‭approach‬ ‭of‬ ‭deep‬ ‭learning‬ ‭is‬ ‭used‬ ‭to‬ ‭train‬ ‭an‬ ‭AI‬ ‭algorithm‬ ‭that‬
‭analyzes‬ ‭chest‬ ‭X‬ ‭ray‬ ‭images‬ ‭and‬ ‭detects‬ ‭the‬ ‭lung‬ ‭disease‬ ‭whether‬ ‭it‬ ‭is‬ ‭pneumonia‬ ‭or‬
‭Tuberculosis‬‭or‬‭Normal.‬‭Deep‬‭Learning‬‭techniques‬‭like‬‭Convolutional‬‭Neural‬‭Networks(CNNs),‬
‭have‬ ‭been‬ ‭successfully‬ ‭employed‬ ‭for‬ ‭image‬ ‭classification‬ ‭in‬ ‭a‬ ‭broad‬ ‭range‬ ‭of‬ ‭medical‬
‭applications‬ ‭(e.g.‬ ‭for‬ ‭pathology‬ ‭detection‬ ‭in‬ ‭X-ray‬ ‭images).‬ ‭CNNs‬ ‭are‬ ‭very‬ ‭similar‬ ‭to‬ ‭normal‬
‭Neural‬ ‭Networks‬ ‭(NNs).‬ ‭They‬ ‭are‬ ‭made‬ ‭up‬ ‭of‬ ‭neurons‬ ‭with‬ ‭their‬ ‭weights‬ ‭and‬ ‭biases.‬ ‭Each‬
‭neuron‬ ‭receives‬ ‭inputs,‬ ‭which‬ ‭outperforms‬ ‭a‬ ‭dot‬ ‭product‬ ‭followed‬ ‭by‬ ‭non-linearity.‬ ‭The‬ ‭loss‬
‭function‬ ‭(Sigmoid)‬ ‭is‬ ‭included‬ ‭in‬ ‭the‬ ‭last‬ ‭layer.‬ ‭These‬ ‭are‬ ‭done‬ ‭at‬ ‭image‬ ‭analysis‬ ‭which‬ ‭is‬
‭designed‬ ‭to‬ ‭process‬ ‭2D‬ ‭inputs.‬ ‭It‬ ‭is‬ ‭also‬ ‭used‬ ‭to‬ ‭learn,‬‭to‬‭detect‬‭spatial‬‭patterns‬‭in‬‭the‬‭training‬
‭data‬‭and‬‭uses‬‭it‬‭to‬‭make‬‭predictions‬‭on‬‭testing‬‭data.‬‭A‬‭model‬‭which‬‭can‬‭predict‬‭diseases‬‭based‬‭on‬
‭X-rays‬‭will‬‭provide‬‭a‬‭reasonable‬‭check‬‭to‬‭help‬‭achieve‬‭higher‬‭accurate‬‭diagnoses.The‬‭network‬‭is‬
‭trained on a large-scale dataset of 3600 X-ray images in order to predict lung diseases.‬
‭1.2 OBJECTIVE OF THE PROJECT‬

‭ bjective‬‭is‬‭to‬‭predict‬‭respiratory‬‭diseases‬‭from‬‭chest‬
O
‭ -ray images using Deep Learning Algorithms like Convolution Neural Network.‬
X

‭1.3 SCOPE OF THE PROJECT‬

‭ espiratory‬ ‭diseases‬‭are‬‭one‬‭of‬‭the‬‭leading‬‭causes‬
R
‭of‬ ‭death.‬ ‭The‬ ‭scope‬ ‭of‬ ‭this‬ ‭project‬ ‭is‬ ‭to‬ ‭assist‬ ‭the‬ ‭radiologists‬ ‭by‬ ‭minimizing‬ ‭the‬ ‭time‬ ‭for‬
‭diagnosing‬‭the‬‭X‬‭rays.The‬‭inherent‬‭modeling‬‭and‬‭efficient‬‭predictive‬‭capabilities‬‭of‬‭Respiratory‬
‭disease‬‭prediction‬‭system‬‭is‬‭explored‬‭to‬‭assist‬‭medical‬‭researchers‬‭in‬‭their‬‭effort‬‭to‬‭evaluate‬‭and‬
‭predict‬‭risk‬‭of‬‭lung‬‭diseases.The‬‭model‬‭is‬‭implemented‬‭using‬‭deep‬‭learning‬‭which‬‭uses‬‭CNN‬‭for‬
‭classification.‬

‭a b c‬

‭Figure 1.1: (a) shows Normal (b) shows Pneumonia and (c) shows Tuberculosis‬
 ‭LITERATURE SURVEY‬

2‭ .1‬‭TITLE:‬‭Deep‬‭Convolutional‬‭Pixel-wise‬‭Labeling‬‭for‬‭lung‬‭Image‬‭Segmentation.‬‭AUTHOR:‬
‭Alehouse,‬ ‭Domenico‬ ‭D.‬ ‭Blois‬ ‭,Mario‬ ‭Muscio‬ ‭,Andrea‬ ‭Pennisi‬ ‭,‬ ‭Daniele‬ ‭Nardi‬ ‭YEAR‬ ‭OF‬
‭PUBLICATION:‬ ‭2018‬ ‭conducted‬ ‭an‬ ‭experiment‬ ‭by‬ ‭comparing‬ ‭public‬‭data‬‭across‬‭two‬‭network‬
‭architectures‬ ‭and‬ ‭using‬ ‭the‬ ‭ISIC‬ ‭database‬ ‭to‬ ‭train‬ ‭the‬ ‭network‬ ‭and‬ ‭the‬ ‭PH2‬ ‭database‬ ‭to‬
‭demonstrate‬‭the‬‭used‬‭method‬‭is‬‭not‬‭data‬‭set‬‭related.‬‭The‬‭results‬‭of‬‭this‬‭experiment‬‭show‬‭that‬‭the‬
‭proposed‬ ‭method‬ ‭is‬ ‭very‬ ‭successful,‬ ‭and‬ ‭even‬ ‭in‬ ‭the‬ ‭presence‬ ‭of‬ ‭hair‬ ‭and‬ ‭air‬ ‭/‬ ‭oil‬ ‭bubbles‬
‭performs‬‭segmentation.‬‭The‬‭extra‬‭feature‬‭of‬‭this‬‭paper‬‭is‬‭data‬‭annotation‬‭which‬‭generates‬‭many‬
‭test images, and the implementation of a semi-automatic Interface.‬

2‭ .2‬ ‭TITLE:‬ ‭Early‬ ‭Detection‬ ‭of‬ ‭Melanoma‬ ‭Lung‬ ‭Cancer‬ ‭Using‬ ‭Classifiers‬ ‭AUTHOR:‬
‭VS.Sabeera,‬ ‭P.Vamsi‬ ‭Krishna‬ ‭YEAR‬ ‭OF‬ ‭PUBLICATION‬ ‭:2016‬ ‭proposed‬ ‭a‬ ‭prevention‬ ‭of‬
‭melanoma‬ ‭and‬ ‭early‬ ‭detection‬ ‭of‬ ‭lung‬ ‭cancer.‬ ‭A‬ ‭test‬ ‭method‬ ‭to‬ ‭assess‬ ‭lesions‬ ‭and‬ ‭to‬ ‭avoid‬
‭melanoma‬‭is‬‭suggested.‬‭A‬‭real-‬‭time‬‭device‬‭has‬‭been‬‭built‬‭to‬‭prevent‬‭people‬‭from‬‭skin‬‭burns‬‭that‬
‭warns‬ ‭users.‬ ‭User‬ ‭receives‬ ‭the‬ ‭warning‬ ‭message‬ ‭when‬ ‭he/she‬ ‭is‬ ‭exposed‬ ‭to‬ ‭sun‬ ‭for‬ ‭a‬ ‭longer‬
‭period and the rays may harm the skin.‬

2‭ .3‬‭TITLE:‬‭Lung‬‭Cancer‬‭Detection‬‭and‬‭classification‬‭AUTHOR:‬‭Pratik‬‭Dubal,‬‭Sankirtan‬‭Bhatt,‬
‭Chaitanya‬ ‭Joglekar,‬ ‭Dr.‬ ‭Sonali‬ ‭Patil‬ ‭10‬ ‭YEAR‬ ‭OF‬ ‭PUBLICATION:‬ ‭2017‬‭Proposed‬‭a‬‭program‬
‭which‬ ‭recognizes‬ ‭and‬ ‭detects‬ ‭lung‬‭cancer‬‭based‬‭on‬‭images‬‭of‬‭general‬‭cameras.‬‭The‬‭images‬‭are‬
‭portioned;‬‭Neural‬‭Network‬‭is‬‭able‬‭to‬‭predict‬‭the‬‭injuries‬‭with‬‭highest‬‭accuracy.‬‭On‬‭a‬‭data‬‭set‬‭of‬
‭463 images, the optimized Neural Network achieved a general accuracy of 76.9%.‬

2‭ .4‬‭TITLE:‬‭Diagnosis‬‭of‬‭Diabetes‬‭Using‬‭Classification‬‭Mining‬‭Techniques‬‭AUTHOR:‬‭Iyer,‬‭J.‬‭S‬
‭R.‬ ‭Sumbaly,‬ ‭YEAR‬ ‭OF‬ ‭PUBLICATION‬ ‭:2016‬ ‭Used‬ ‭the‬ ‭Decision‬ ‭Tree‬ ‭and‬ ‭Naïve‬ ‭Bayes‬
‭algorithms,‬‭the‬‭goal‬‭is‬‭to‬‭find‬‭solutions‬‭for‬‭diabetes‬‭detection‬‭by‬‭analyzing‬‭patterns‬‭originating‬‭in‬
‭the‬ ‭data‬ ‭through‬ ‭classification‬ ‭analysis.‬ ‭This‬ ‭becomes‬ ‭an‬ ‭effective‬ ‭method‬ ‭of‬ ‭detecting‬ ‭the‬
‭disease‬ ‭that‬ ‭will‬‭help‬‭patients‬‭recover‬‭well‬‭in‬‭time.‬‭The‬‭study‬‭concluded‬‭that‬‭the‬‭J48‬‭algorithm‬
‭provides an accuracy rate of 74.8 percent.‬

2‭ .5‬ ‭TITLE:‬ ‭Comparison‬ ‭of‬ ‭Feature‬ ‭selection‬ ‭methods‬ ‭for‬ ‭diagnosis‬ ‭of‬ ‭cervical‬ ‭cancer‬ ‭using‬
‭Support‬ ‭vector‬ ‭machine‬ ‭(SVM)‬ ‭classifier.‬ ‭AUTHOR:‬ ‭B.‬ ‭Ashok‬ ‭and‬ ‭P.‬ ‭Aruna,‬ ‭YEAR‬ ‭OF‬
‭PUBLICATION:‬‭2016‬‭proposed‬‭Comparison‬‭of‬‭feature‬‭selection‬‭approaches‬‭for‬‭cervical‬‭cancer‬
‭diagnosis‬ ‭using‬ ‭a‬ ‭SVM‬ ‭classifier.‬ ‭Image‬ ‭segmentation‬ ‭was‬ ‭performed‬ ‭using‬ ‭thresholding.‬
‭Feature‬ ‭selection‬ ‭was‬ ‭attained‬ ‭using‬ ‭mutual‬ ‭information,‬ ‭sequential‬ ‭forward‬ ‭search,‬‭sequential‬
‭floating forward search, and random subset feature selection methods.‬
 ‭PROPOSED METHODOLOGY‬

‭3.1 DATASET DESCRIPTION‬

‭ he‬ ‭dataset‬ ‭used‬ ‭is‬ ‭amassed‬ ‭by‬ ‭the‬ ‭NIH‬ ‭(National‬ ‭Institutes‬ ‭of‬ ‭Health‬ ‭-‬ ‭Clinical‬ ‭Center‬ ‭).‬ ‭It‬
T
‭comprises‬‭chest‬‭X-ray‬‭images‬‭of‬‭5860‬‭from‬‭1356‬‭unique‬‭patients.‬‭The‬‭dataset‬‭is‬‭available‬‭from‬
‭open‬‭source.‬‭3600‬‭of‬‭images‬‭are‬‭used‬‭here.‬‭Each‬‭image‬‭in‬‭the‬‭dataset‬‭is‬‭labeled‬‭with‬‭Pneumonia‬
‭and‬‭Tuberculosis‬‭and‬‭Normal.‬‭The‬‭deep‬‭learning‬‭model‬‭uses‬‭224x224‬‭resolution‬‭and‬‭normalizes‬
‭the‬ ‭data‬ ‭subsequently.‬ ‭Each‬ ‭pixel‬ ‭has‬ ‭zero‬ ‭mean‬ ‭and‬ ‭approximately‬ ‭unit‬ ‭variance‬ ‭and‬
‭downsampled‬‭in‬‭the‬‭neural‬‭net‬‭to‬‭predict‬‭disease‬‭as‬‭either‬‭Pneumonia‬‭or‬‭Tuberculosis‬‭or‬‭Normal.‬
‭The below fig shows the example of the input image.‬

‭Figure 3.1: Example of Lung X-ray image as input‬

I‭ n‬‭this‬‭proposed‬‭work‬‭we‬‭try‬‭to‬‭design‬‭a‬‭model‬‭which‬‭can‬‭be‬‭used‬‭for‬‭prediction‬‭of‬‭lung‬‭cancer‬
‭from‬ ‭real‬ ‭world‬ ‭chest‬ ‭x-ray‬ ‭images.‬ ‭For‬ ‭training‬ ‭the‬ ‭system‬ ‭we‬‭try‬‭to‬‭collect‬‭the‬‭sample‬‭chest‬
‭X-ray‬‭images‬‭from‬‭KAGGLE‬‭website‬‭and‬‭then‬‭train‬‭the‬‭system.‬‭Once‬‭the‬‭system‬‭is‬‭trained‬‭now‬
‭we‬ ‭can‬ ‭check‬ ‭the‬ ‭model‬‭performance‬‭by‬‭giving‬‭dynamic‬‭images‬‭and‬‭check‬‭the‬‭performance‬‭of‬
‭each‬ ‭and‬ ‭every‬ ‭individual‬ ‭Model.‬ ‭Using‬ ‭Deep‬ ‭Learning‬ ‭to‬ ‭predict‬ ‭lung‬ ‭diseases‬ ‭from‬ ‭Chest‬
‭X-rays‬‭can‬‭be‬‭a‬‭lifesaving‬‭factor‬‭for‬‭an‬‭individual‬‭suffering‬‭from‬‭the‬‭disease.‬‭This‬‭is‬‭possible‬‭as‬
‭the‬‭results‬‭can‬‭be‬‭predicted‬‭with‬‭a‬‭high‬‭percentage‬‭of‬‭accuracy‬‭instantly.‬‭This‬‭project‬‭presents‬‭an‬
‭effective‬‭way‬‭for‬‭expert‬‭diagnosis‬‭of‬‭lung‬‭diseases‬‭using‬‭Deep‬‭Learning.‬‭It‬‭focuses‬‭on‬‭creating‬‭a‬
‭system‬ ‭for‬ ‭assistance‬ ‭of‬ ‭Radiologists‬ ‭in‬ ‭detection‬ ‭of‬‭lung‬‭diseases.‬‭This‬‭will‬‭especially‬‭benefit‬
‭rural areas where radiologists aren’t easily available.‬

‭ADVANTAGES OF THE PROPOSED SYSTEM‬

1‭ . The proposed scheme is very accurate in classification of check x–ray images‬

‭2. The proposed system gives accurate recommendations for the doctors.‬
‭3.‬‭The‬‭proposed‬‭system‬‭is‬‭capable‬‭of‬‭classification‬‭of‬‭chest‬‭x-ray‬‭and‬‭finding‬‭out‬‭the‬‭accuracy‬
‭of images.‬
I‭ n‬ ‭this‬‭section‬‭we‬‭try‬‭to‬‭discuss‬‭the‬‭proposed‬‭CNN‬‭model‬‭which‬‭is‬‭used‬‭to‬‭detect‬‭chest‬‭disease‬
‭using pre-trained CNN models.‬
‭The Application is mainly divided into 4 modules.They are as follows:‬
‭1. Convolution Layer‬
‭2. Rectified Linear Unit (RELU) Layer‬
‭3. Pooling Layer‬
‭4. Fully Connected layer‬

‭A) CONVOLUTION LAYER‬

‭ ‬ ‭convolution‬ ‭is‬ ‭defined‬ ‭as‬ ‭an‬ ‭operation‬ ‭on‬ ‭two‬ ‭functions.‬ ‭In‬ ‭image‬ ‭analysis,‬ ‭one‬ ‭function‬
A
‭consists‬‭of‬‭input‬‭values‬‭(e.g.‬‭pixel‬‭values)‬‭at‬‭a‬‭position‬‭in‬‭the‬‭image,‬‭and‬‭the‬‭second‬‭function‬‭is‬‭a‬
‭filter‬ ‭(or‬ ‭kernel)‬ ‭each‬ ‭can‬ ‭be‬ ‭represented‬ ‭as‬ ‭an‬ ‭array‬ ‭of‬ ‭numbers.‬ ‭Computing‬ ‭the‬ ‭dot‬ ‭product‬
‭between‬ ‭the‬ ‭two‬ ‭functions‬ ‭gives‬ ‭an‬ ‭output.‬ ‭The‬ ‭filter‬‭is‬‭then‬‭shifted‬‭to‬‭the‬‭next‬‭position‬‭in‬‭the‬
‭image‬ ‭as‬ ‭defined‬ ‭by‬ ‭the‬ ‭stride‬ ‭length.‬ ‭The‬ ‭computation‬ ‭is‬ ‭repeated‬ ‭until‬ ‭the‬ ‭entire‬ ‭image‬ ‭is‬
‭covered,‬ ‭producing‬ ‭a‬ ‭feature‬ ‭(or‬ ‭activation)‬ ‭map.‬ ‭This‬ ‭is‬ ‭a‬ ‭map‬ ‭of‬ ‭where‬ ‭the‬ ‭filter‬‭is‬‭strongly‬
‭activated‬‭and‬‭‘sees’‬‭a‬‭feature‬‭such‬‭as‬‭a‬‭straight‬‭line,‬‭a‬‭dot,‬‭or‬‭a‬‭curved‬‭edge.‬‭If‬‭a‬‭photograph‬‭of‬‭a‬
‭face‬ ‭was‬ ‭fed‬ ‭into‬ ‭a‬‭CNN,‬‭initially‬‭low-level‬‭features‬‭such‬‭as‬‭lines‬‭and‬‭edges‬‭are‬‭discovered‬‭by‬
‭the‬ ‭filters.‬ ‭These‬‭build‬‭up‬‭to‬‭progressively‬‭higher‬‭features‬‭in‬‭subsequent‬‭layers,‬‭such‬‭as‬‭a‬‭nose,‬
‭eye or ear, as the feature maps become inputs for the next layer in the CNN architecture.‬

‭B) RECTIFIED LINEAR UNIT (RELU) LAYER‬

‭ he‬‭RELU‬‭layer‬‭is‬‭an‬‭activation‬‭function‬‭that‬‭sets‬‭negative‬‭input‬‭values‬‭to‬‭zero.‬‭This‬‭simplifies‬
T
‭and‬ ‭accelerates‬ ‭calculations‬ ‭and‬ ‭training,‬ ‭and‬ ‭helps‬ ‭to‬ ‭avoid‬ ‭the‬ ‭vanishing‬ ‭gradient‬ ‭problem.‬
‭Mathematically‬ ‭it‬ ‭is‬ ‭defined‬ ‭as:‬ ‭f(x)‬ ‭=max‬ ‭(0,‬ ‭x).‬ ‭Where‬ ‭x‬ ‭is‬ ‭the‬ ‭input‬ ‭to‬ ‭the‬ ‭neuron.‬ ‭Other‬
‭activation‬ ‭functions‬ ‭include‬ ‭the‬ ‭sigmoid,‬ ‭tanh,‬ ‭leaky‬ ‭RELUs,‬ ‭Randomized‬ ‭RELUs‬ ‭and‬
‭parametric RELUs.‬

‭C) POOLING LAYER‬

‭ he‬‭Pooling‬‭layer‬‭is‬‭inserted‬‭between‬‭the‬‭Convolution‬‭and‬‭RELU‬‭layers‬‭to‬‭reduce‬‭the‬‭number‬‭of‬
T
‭parameters‬ ‭to‬ ‭be‬ ‭calculated,‬ ‭as‬ ‭well‬ ‭as‬ ‭the‬‭size‬‭of‬‭the‬‭image‬‭(width‬‭and‬‭height,‬‭but‬‭not‬‭depth).‬
‭Max-pooling‬ ‭is‬ ‭most‬ ‭commonly‬ ‭used;‬ ‭other‬ ‭pooling‬ ‭layers‬ ‭include‬ ‭Average‬ ‭pooling‬ ‭and‬
‭L2-normalization‬ ‭pooling.‬ ‭Maxpooling‬ ‭simply‬ ‭takes‬ ‭the‬ ‭largest‬ ‭input‬ ‭value‬ ‭within‬ ‭a‬ ‭filter‬ ‭and‬
‭discards‬ ‭the‬ ‭other‬ ‭values;‬ ‭effectively‬ ‭it‬ ‭summarizes‬ ‭the‬ ‭strongest‬ ‭activations‬ ‭over‬ ‭a‬
‭neighborhood.‬‭The‬‭rationale‬‭is‬‭that‬‭the‬‭relative‬‭location‬‭of‬‭a‬‭strongly‬‭activated‬‭feature‬‭to‬‭another‬
‭is more important than its exact location.‬

‭D) FULLY CONNECTED LAYER‬

 ‭ he‬ ‭final‬ ‭layer‬ ‭in‬ ‭a‬ ‭CNN‬ ‭is‬ ‭the‬ ‭Fully‬ ‭Connected‬ ‭Layer,‬ ‭meaning‬ ‭that‬ ‭every‬ ‭neuron‬ ‭in‬ ‭the‬
T
‭preceding‬‭layer‬‭is‬‭connected‬‭to‬‭every‬‭neuron‬‭in‬‭the‬‭Fully‬‭Connected‬‭Layer.‬‭Like‬‭the‬‭convolution,‬
‭RELU‬‭and‬‭pooling‬‭layers,‬‭there‬‭can‬‭be‬‭1‬‭or‬‭more‬‭fully‬‭connected‬‭layers‬‭depending‬‭on‬‭the‬‭level‬
‭of‬‭feature‬‭abstraction‬‭desired.‬‭This‬‭layer‬‭takes‬‭the‬‭output‬‭from‬‭the‬‭preceding‬‭layer‬‭(Convolution,‬
‭RELU‬ ‭or‬ ‭Pooling)‬ ‭as‬ ‭its‬ ‭input,‬ ‭and‬ ‭computes‬ ‭a‬ ‭probability‬ ‭score‬ ‭for‬ ‭classification‬ ‭into‬ ‭the‬
‭different‬ ‭available‬ ‭classes.‬ ‭In‬ ‭essence,‬ ‭this‬ ‭layer‬ ‭looks‬ ‭at‬ ‭the‬‭combination‬‭of‬‭the‬‭most‬‭strongly‬
‭activated‬ ‭features‬ ‭that‬ ‭would‬ ‭indicate‬ ‭the‬ ‭image‬ ‭belongs‬ ‭to‬ ‭a‬ ‭particular‬‭class.‬‭For‬‭example,‬‭on‬
‭histology‬ ‭glass‬ ‭slides,‬ ‭cancer‬ ‭cells‬ ‭have‬ ‭a‬ ‭high‬ ‭DNA‬ ‭to‬ ‭cytoplasm‬ ‭ratio‬ ‭compared‬ ‭to‬ ‭normal‬
‭cells.‬
 ‭RESULTS‬

‭SOFTWARE RESULTS‬
 ‭CONCLUSION‬

‭In‬‭conclusion,‬‭we‬‭for‬‭the‬‭first‬‭time‬‭designed‬‭and‬‭implemented‬‭an‬‭application‬‭using‬‭deep‬‭learning‬
‭CNN‬‭model‬‭in‬‭the‬‭medical‬‭field‬‭for‬‭Respiratory‬‭disease‬‭detection‬‭from‬‭chest‬‭X-ray‬‭images.Our‬
‭project‬ ‭uses‬ ‭clever‬ ‭technology‬ ‭to‬ ‭quickly‬‭find‬‭if‬‭someone‬‭might‬‭have‬‭lung‬‭sickness‬‭from‬‭chest‬
‭X-ray‬‭pictures.‬‭By‬‭teaming‬‭up‬‭with‬‭a‬‭sensor-equipped‬‭device,‬‭we‬‭not‬‭only‬‭check‬‭the‬‭air‬‭around‬‭a‬
‭person‬‭but‬‭also‬‭keep‬‭an‬‭eye‬‭on‬‭their‬‭temperature.‬‭If‬‭the‬‭computer‬‭thinks‬‭someone‬‭could‬‭be‬‭sick,‬
‭this device helps us decide if it's safe for them to go outside.‬

‭The‬‭best‬‭part‬‭is‬‭that‬‭if‬‭someone‬‭needs‬‭to‬‭stay‬‭indoors,‬‭we‬‭make‬‭sure‬‭they‬‭get‬‭a‬‭clear‬‭voice‬‭alert.‬
‭This‬‭way,‬‭everyone‬‭stays‬‭well-informed‬‭about‬‭their‬‭health,‬‭and‬‭we‬‭all‬‭work‬‭together‬‭to‬‭keep‬‭each‬
‭other‬ ‭safe.‬ ‭Our‬ ‭goal‬ ‭is‬ ‭to‬ ‭make‬ ‭things‬ ‭easier‬ ‭for‬ ‭doctors,‬ ‭help‬ ‭people‬ ‭get‬ ‭care‬ ‭sooner,‬ ‭and‬
‭contribute‬‭to‬‭the‬‭health‬‭of‬‭the‬‭whole‬‭community.‬‭This‬‭project‬‭shows‬‭how‬‭smart‬‭tech‬‭and‬‭a‬‭bit‬‭of‬
‭care‬ ‭can‬ ‭make‬ ‭a‬ ‭big‬ ‭difference‬ ‭in‬ ‭looking‬ ‭after‬ ‭each‬ ‭other's‬ ‭well-being.Future‬ ‭works‬ ‭on‬
‭additional‬‭data‬‭collections‬‭and‬‭validation‬‭will‬‭further‬‭improve‬‭the‬‭generalizability‬‭of‬‭the‬‭current‬
‭methods.‬ ‭Respiratory‬ ‭disease‬ ‭is‬ ‭an‬ ‭extremely‬ ‭complex‬ ‭problem‬ ‭to‬ ‭solve‬ ‭however‬ ‭with‬ ‭early‬
‭detection a patient has a high increase of survivability.‬
 ‭REFERENCES‬

‭[1]Deep Convolutional Pixel-wise Labeling for Lungs Image Segmentation, AL Youssef;‬

‭Domenico D. Bloisi ;Mario Muscio ; Andrea Pennisi ; Daniele Nardi ; Antonio Facchiano IEEE‬
‭International Symposium on Medical Measurements and Applications (MeMeA) Year: 2018‬
‭Patchoo, W. 2018 International Workshop on Advanced Image Technology.‬

‭[2] Early Detection of Melanoma Lung Diseases Using Classifiers, VS. Sabeera, P.‬
‭VamsiKrishna PG Scholar,Dept.ofECE,R.K.Collegeof Engineering, Vijayawada, Andhra Pradesh‬
‭Assistant Professor, Dept. of ECE, R.K. College of Engineering, Vijayawada, Andhra Pradesh,‬
‭Year2016‬

‭[3] Vembandasamy, K., Sasipriya, R. and Deepa, E. (2015) Heart Diseases Detection Using‬
‭Naive Bayes Algorithm. IJISET- International Journal of Innovative Science, Engineering‬
‭&Technology.‬

‭[4] Lung Cancer Detection and classification Pratik Dubal, Sankirtan Bhatt, Chaitanya Joglekar,‬
‭Dr. Sonali Patil, Department of Information Technology K. J. Somaiya College of Engineering‬
‭Vidyavihar,Year-2017. [5] Iyer, J. S and R. Sumbaly, "Diagnosis of Diabetes Using Classification‬
‭Mining Techniques", IJDKP, vol.5, no. 1, pp. 01- 14,2015.‬

‭[6] B. Ashok and P. Aruna, “Comparison of Feature selection 30 methods for diagnosis of‬
‭cervical cancer using support vector machine(SVM) classifier,” vol. 6, no. 1, pp. 94–99, 2016.‬

‭[7] X. Pan, Y. Luo, Y. Xu, “K-nearest neighbour based structural twin support vector machine,”‬
‭Knowledge Based Systems, vol. 88, pp. 34-44, 2015.‬

‭[8] Vijayarani, S. and Dhayanand, S. (2015) Liver Disease Prediction using SVM and Naïve‬
‭Bayes Algorithms. International Journal of Science, Engineering and Technology Research‬
‭(IJSETR), 4, 816-820.‬