FORM - 2 THE PATENTS ACT, 1970 (39 of 1970) COMPLETE SPECIFICATION (Section 10, rule 13) DETECTION OF MALARIA PARASITES USING MACHINE LEARNING NAME OF THE APPLICANTS: KHUSHBU WAMAN KAWALE MASUMI BHAVESH SHAH Residing at Plot No. B, Sector No. 110, Gate No. 1, Laxminagar, Ravet, Haveli, Pune, Maharashtra, 412101 FIELD OF THE INVENTION: This invention is related to field of image processing along with CNN based DL model. Particularly, this invention relates tothe performance of pre-trained CNN based DL model as feature extractors towards classifying the parasitized and uninfected cells to aid in improve disease training. The important contribution of this work are:(a) presentation of comparative analysis of performance of customized and pre-trained DL models as feature extractors toward classifying parasitized and uninfected cells, (b) cross validating the performance of predictive models at the patient level to reduce bias and generalization errors, (c) analysis and selection of optimal layers in pre-trained models to extract features from underlying data and (d) testing for the presence/absence of statistically significant difference in performance of customized and pre-trained CNN models under study. BACKGROUND OF THE INVENTION: Malaria is a mosquito-borne blood disease caused by plasmodium parasites transmitted through the bite of female Anopheles mosquito. There are several methods and tests which can be used for malaria detection and diagnosis. Based on the guidelines from the WHO protocol, this procedure involves intensive examination of the blood smear at a 100X magnification, where people manually count red blood cells that contain parasites out of 5000 cells.The diagnostic accuracy heavily depends on human expertise, a clinician may have to manually count up to 5,000 cells, an extremely tedious and time-consuming process. In order to help make malaria testing a faster process in the field, scientists and researchers have developed antigen tests for Rapid Diagnosis Testing (RDT). It uses a small device that allows both a blood sample and a buffer to be added. Internally, the device performs the test and provides the results. While RDTs are significantly faster than cell counting they are also much less accurate. With regular manual diagnosis of blood smears, it is an intensive manual process requiring proper expertise in classifying and counting the parasitized and uninfected cells. Typically, this may not scale well and might cause problems if we do not have the right expertise in specific regions around the world. So, we have made some advancement in image processing and analysis techniques to extract hand-engineered features and build machine learning based classification models. PRIOR ART: — According to prior art it is observed that the diagnostic accuracy heavily depends on human expertise and can be adversely impacted by the inter observer variability and liability imposed by large scale diagnosis in disease endemic regions. Alternating techniques such as Polymerase Chain Reaction (PCR) and Rapid Diagnostic Tests (RDT) are used, however, PCR analysis is limited in its performance and RDT is less cost effective in disease endemic regions. To overcome challenges of devising hand-engineered features that capture variations in the underlying data, pre trained DL models are used as feature extractors to aid in visual recognition tasks. CNN's trained on large scale dataset serve as feature extractors for wide range of computer vision tasks to improve performance. OBJECTIVE OF THE INVENTION: Malaria is a mosquito-borne blood disease caused by plasmodium parasites transmitted through the bite of female Anopheles mosquito. In 2016, World Health Organization reported 212 million instances of disease across the world.Nearly half the world's population is at risk from malaria and there are over 200 million malaria casesand approximately 400,000 deaths due to malaria every year. This gives us all more motivation to makemalaria detection and diagnosis fast, easy and effective. There are several methods and tests which canbe used for malaria detection and diagnosis.Microscopic thick and thin bloodsmear examinations are the most reliable and commonly used method for disease diagnosis.Thick blood smears assist in detecting the presence of parasites while thin blood smears assist in identifying the species of the parasite causing the infection. The diagnostic accuracyheavily depends on human expertise and can be adversely impacted by the inter observervariability and liability imposed by large-scale diagnosis in disease endemic regions. To overcome challenges of devising hand-engineered features that capture variations in theunderlying data, Deep Learning (DL) has been used with significant success. SUMMARY OF INVENTION: For images, an important source of information lies in the spatial local correlation amongthe neighboring pixels. Convolutional Neural Networks, a class of DL models is designedto exploit this information through the mechanism of local receptive fields, shared weightsand pooling. According to this invention, We will segment the red blood cells from thin blood smear images andrandomly split into train and test sets. A total of 25% of training images were randomlyselected to validate the models. Images are then re-sampled to 50x50 pixel resolution to compensate for lack of computational resources. Once the acquired images are resized, we will perform preprocessing tasks on the images like Image Resizing, Grayscale Conversion and Image Thresholding to convert it to binary format so that the processing of the image becomes easier while putting it in the pooling layers of the CNN. In this work, we evaluated the performance of pre-trained CNN based DL model asfeatureextractors towards classifying the parasitized and uninfected cells to aid in improve disease training. DETAILED DESCRIPTION OF THE ACCOMPANYING DRAWINGS: In the figure 100, the number indicates, Reference numeral 102, relates to Image Acquisition Reference numeral 104, relates to Image Pre-processing Reference numeral 106, relates to Feature Extraction Reference numeral 108, relates to Thresholding Reference numeral 110, relates to Comparison of Database with Pre-trained model Reference numeral 112, relates to Result TERMINOLOGY Polymerase Chain Reaction (PCR) It is a technique to make many copies of a specific DNA region in vitro (in a test tube rather than an organism). Rapid Diagnostic Test (RDT) A rapid diagnostic test is a medical diagnostic test that is quick and easy to perform. RDTs are suitable for preliminary or emergency medical screening. Convolutional Neural Network (CNN) It is a specific type of artificial neural network for supervised learning and to analyze data. K Nearest Neighbors (k-NN) It is an algorithm which belongs to the supervised learning domain and finds intense application in pattern recognition, data mining and intrusion detection. OpenCV OpenCV (Open Source Computer Vision Library) is an open source computer vision and machinelearning software library. Keras .Keras is a high-level neural networks API, written in Python and capable of running on top of Tensor Flow, CNTK, or Theano. Spyder Spyder has great recognition in the IDE market. It is the most suitable Python IDE for data science works. Jupyter Lab/Notebook The Jupyter Notebook is an open-source web application that allows you to create and share documents that contain live code, equations, visualizations and narrative text. Image Acquisition It is defined as the action of retrieving an image from some source, usually a hardware-based source. Image Pre-processing It is an improvement of the image data that suppresses unwanted distortions or enhances some image features important for further processing. Feature Extraction It is a process which starts from an initial set of measured data and builds derived values intended to be informative and non-redundant which facilitates the subsequent learning and generalization. Thresholding It is an image processing method which is based on setting a threshold value on the pixel intensity of the original image. It is a way of partitioning an image into a foreground and background. DETAILED DESCRIPTION OF THE INVENTION: The system comprises of the use of several image processing methods. Machine Learning algorithms such as CNN and k-NN are used for distinguishing between the healthy and infected blood samples. The system also provides a comparative analysis between the algorithms used for the task. The images form the available dataset will be taken and then the pre-processing steps will be applied on the available image such as resizing the images, noise removal, angle correction, brightness correction, etc. These processed images will be further converted into gray scale images and then the converted will be binarized. These binarized images has two values i.e. 0 and 1, which will be used in the feature extraction process to identify the spots in the blood smear images. The thresholding process will be used to segment the images into individual images. After all these processes are done the model will be trained. As we have a pre-trained model in which the images have gone under some processing.Furthermore, the test images will be compared with the trained model and then segregated into the infected and healthy blood images.