Scribd
Upload
6 views

05_AIHC_Exp03

Uploaded by

laxitac115
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
6 views

05_AIHC_Exp03

Uploaded by

laxitac115
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 7

Vidyavardhini’s College of Engineering & Technology

Name: Durvesh Kajrekar


Class: BE/CSE-DS
Experiment No. 3

Perform AI for medical diagnosis based on MRI/X-ray data


using CNN
Date of Performance: 9/8/24
Date of Submission: 23/8/24

HAIMLSBL701 AI&ML in Healthcare Lab


Vidyavardhini’s College of Engineering & Technology

Aim: Perform AI for medical diagnosis based on MRI/X-ray data using CNNa.

Objective: The objective of this experiment is to use Convolutional Neural Networks to


identify Pneumonia from the dataset of Xray images.

Theory:
Convolutional Neural Networks (CNNs), also known as ConvNets, are a class of deep
learning neural networks specifically designed for tasks related to computer vision, such as
image classification, object detection, image segmentation, and more. They have had a
profound impact on various fields, including image and video analysis, medical imaging,
autonomous vehicles, and many others. CNNs were inspired by the human visual system and
are highly effective at processing grid-like data, making them particularly suited for tasks
involving images and spatial data.

Here's an introduction to the key concepts and components of Convolutional


Neural Networks:

Convolutional Layers (Convolution):


Convolution is the fundamental operation in CNNs. It involves sliding a small filter (also
called a kernel) over the input data to perform element-wise multiplication and summation.
This process helps detect features like edges, textures, and patterns within the data.
Convolutional layers consist of multiple filters that learn different features from the input
data. These layers can capture increasingly complex features as you move deeper into the
network.

Pooling Layers (Subsampling or Pooling):


Pooling layers are used to downsample the spatial dimensions of the feature maps produced
by convolutional layers. Common pooling operations include max-pooling and average-
pooling. Pooling reduces the computational burden and helps make the network more robust
to small translations and distortions in the input.

Activation Functions:
Activation functions like ReLU (Rectified Linear Unit) are applied to the output of
convolutional and pooling layers. ReLU introduces non-linearity into the network, allowing it
to learn complex relationships within the data.

HAIMLSBL701 AI&ML in Healthcare Lab


Vidyavardhini’s College of Engineering & Technology

Fully Connected Layers (FC Layers):


After several convolutional and pooling layers, CNNs often have one or more fully connected
layers, which are traditional neural network layers. These layers perform high-level feature
extraction and map the extracted features to the final output classes in tasks like image
classification.

Flattening:
Before passing data from the convolutional layers to the fully connected layers, the feature
maps are flattened into a one-dimensional vector. This transformation allows the network
to treat the data as a traditional feedforward neural network.

Backpropagation:
CNNs are trained using backpropagation and gradient descent algorithms. During training,
the network adjusts its internal parameters (weights and biases) to minimize a loss function,
which measures the difference between the predicted output and the ground truth labels.

Architectural Variations:
CNN architectures can vary widely in terms of depth, number of layers, and specific
components. Popular CNN architectures include LeNet, AlexNet, VGG, GoogLeNet
(Inception), and ResNet, among others.

Transfer Learning:
CNNs have shown that pre-trained models on large datasets can be fine-tuned for specific
tasks with smaller datasets. This approach, called transfer learning, has become common in
computer vision applications. CNNs have revolutionized the field of computer vision and are
also used in various other domains like natural language processing and reinforcement
learning. They excel at learning hierarchical representations of data, which makes them a
powerful tool for a wide range of machine learning tasks involving structured grid-like data.

HAIMLSBL701 AI&ML in Healthcare Lab


Vidyavardhini’s College of Engineering & Technology

Code and Output:

HAIMLSBL701 AI&ML in Healthcare Lab


Vidyavardhini’s College of Engineering & Technology

HAIMLSBL701 AI&ML in Healthcare Lab


Vidyavardhini’s College of Engineering & Technology

Google Collaboratory Link: -

HAIMLSBL701 AI&ML in Healthcare Lab


Vidyavardhini’s College of Engineering & Technology

https://colab.research.google.com/drive/1ag0lKtPTd-gwZVbprq_bfo4RdFLInbNm?
usp=sharing

Conclusion: -
The CNN architecture is especially useful for image recognition and image classification as
they can effectively extract features from images and learn to recognize patterns, making
them well-suited for tasks such as object detection, image segmentation, and classification.
They can process large amounts of data and produce highly accurate predictions.They can
learn from raw pixel data, without requiring any manual feature engineering or preprocessing.

HAIMLSBL701 AI&ML in Healthcare Lab

You might also like