TOPIC: MIE-IF/AB: MEDICAL IMAGE

ENHANCEMENT USING IMAGE

FUSION AND ALPHA BLENDING

COCHIN UNIVERSITY COLLEGE OF ENGINEERING KUTTANADU

Submitted by :
1. FIDA SAHOODA P - 20320514
2. NAMAN ANAND - 20320522
PROJECT CO ORDINATOR: Dr. ANILKUMAR K K 3. ASWIN K - 20320540
PROJECT GUIDE : Mr. NISHANTH R 4. MAYURI P - 20920518
 CONTENT
1. INTRODUCTION
2. CAUSES AND SOLUTIONS
3. EXISTING TECHNOLOGY
4. METHODOLOGY
5. IMAGE COLLECTION
6. MASKING
7. PRE-PROCESSING
8. IMAGE FUSION AND ALPHA BLENDING
9. QUALITY EVALUATION
10. RESULTS
11. ADVANTAGES AND DISADVANTAGES
12. FUTURE SCOPE
13. CONCLUSION
14. REFERENCES
 INTRODUCTIO
N
• OBJECTIVE : Enhance diagnostic image quality in medical imaging through image
processing techniques.
• Challenges in Medical Imaging: Motion artifacts, processing errors, equipment
malfunctions, and complex patient anatomy compromise diagnostic accuracy.
• Solutions: Meticulous patient positioning, quality control, and optimized imaging
parameters.
• Tools and Technologies: Utilization of MATLAB for various processing tasks.
Integration of machine learning, especially CNNs, for defect detection and correction.
• Impact on Diagnostic Accuracy: Integration of techniques bolsters diagnostic accuracy,
advancing patient care in medical imaging.
• Image Blending and Fusion: Blending for smooth merging, fusion for integrated
information from multiple images.
 CAUSES AND
SOLUTIONS
CAUSES SOLUTIONS

Machine
Inadequate Learning
Imaging Segmentation
Motion Artifacts
Parameter

Image Processing Patients Artifacts

Blending
Errors Anatomy removal

Equipment
Registeration Deconvolution
Malfunction
EXISTING TECHNOLOGY
• Inpainting and interpolation
• Registeration and fusion
• Deep learning based approach
• Patch based approach
 METHODOLOGY
• Collecting Images from DATASET
• Masking of the Required Portion
• Pre processing
 Normalization
Noise removal
Contrast enhancement
Augmentation
resizing
• IMAGE FUSION AND ALPHA BLENDING
• QUALITY ANALYSIS
RMS
PSNR
 SSIM
 Collecting Images from DATASET
STEPS IN COLLECTING IMAGES FROM DATA SET
• Dataset Selection
• Define Criteria
• Visual Inspection
• Validation
• Storage and Organization
• Ethical Considerations

Kenhub : It is an online platform that provides educational resources for medical students and
professionals. It offers anatomy and medical imaging courses, tutorials, quizzes, and interactive
tools to aid in learning human anatomy and related subjects.
Kaggle : It is a platform for data science competitions, datasets, and courses. It hosts machine
learning competitions where companies and researchers can post their data and problem statements,
inviting data scientists and machine learning experts to develop predictive models and solutions.
 Masking of the Required Portion
• In medical image analysis, masking refers to the process of isolating or highlighting
specific regions or structures within an image for further analysis or visualization.
• We have masked the images using binary mask
• A binary mask in image processing is a black-and-white image used to selectively
apply operations on another image. In this mask, pixels marked as white (typically
with a value of 1) indicate areas of interest, while black pixels (usually with a value
of 0) denote areas to be excluded.
 Pre processing
1. NORMALIZATION
Intensity normalization scales pixel values within medical images to a
standardized range, improving consistency for quantitative analysis. It ensures that
pixel intensities are comparable across images acquired with different imaging
parameters or scanners.
• We use min-max normalization method
• Xnorm​=(X−Xmin) /​(Xmax−Xmin​​)
• where:
• Xnorm​is the normalized value,
• X is the original value
• Xmin​is the minimum pixel value
• Xmax​is the maximum pixel value
2.Noise removal
• Noise reduction techniques aim to remove or reduce unwanted artifacts or
distortions present in medical images. Common methods include Gaussian blur,
median filtering, and bilateral filtering.
• Applying a Gaussian filter to an image involves convolving the image with a Gaussian
kernel. This operation is commonly used for noise reduction because it blurs the
image, smoothing out small variations in pixel values caused by noise while
preserving the overall structure and edges of the image. The "sigma" parameter in
the Gaussian filter specifies the standard deviation of the Gaussian distribution,
controlling the extent of blurring.

(a)Original image
(b)Noise reduced image
 3. Contrast Enhancement
• Contrast enhancement techniques improve the visibility of structures within medical
images by adjusting the intensity distribution. Histogram equalization, contrast
stretching, and adaptive histogram equalization are commonly used for this purpose.
• First calculates J = histeq(I, n);
• CDF(k)=∑i=0k​histogram(i)
• It maps the original pixel intensities to new
intensities using the CDF.
• Replaces each pixel's intensity with the
corresponding mapped intensity
Before Equalization:
This histogram might show a peak at a specific intensity value, indicating a large portion of the image has a similar
brightness. There could be limited spread across the entire intensity range, suggesting low contrast between different
tissues.

After Equalization:
This histogram will ideally be more spread out across the entire intensity range. Each intensity value should have a
more balanced representation, indicating improved contrast between different brain regions
 4. Augmentation
Augmentation is a technique used especially in medical image analysis, where
transformations like rotation, flipping, and scaling are applied to images. This helps to
increase the diversity of the dataset. In medical imaging tasks such as classification and
segmentation, where images can vary significantly, augmentation is valuable for training
models effectively.

5. Resizing
Re-sizing adjusts image dimensions to a desired
size, crucial in standardizing resolution and aspect
ratio across data sets
 IMAGE FUSION AND ALPHA BLENDING
• Image Fusion:
• Combining multiple images from different sources or modalities.
• Each image may have its own errors or artifacts.
• Fusion compensates for errors in one image with information from others.
• Example: Combining a noisy image with a blurry one to get a clear, noise-reduced result.
• Image Blending:
• Refines fused images by seamlessly integrating them.
• Techniques like alpha blending or gradient domain blending are used.
• Smooth transitions between images are ensured.
• Maintains consistency across the composite image.
• Helps combine strengths of images while minimizing individual errors.
• Results in visually appealing and accurate representations of the data.
Equation explanation of alpha blending

• blended_ image = uint8((alphas(i) * double(imagel) + alphas(i) •k

double(image2) + alphas(i) * double(image3) + alphas(i) •k
double(image4) + alphas(i) * double(image5)) / 5);

• Blending Parameter (alphas)

• Type Conversion- double()
• Weighted Sum
• Normalization
• Data Type Conversion- uint8()
 QUALITY EVALUTION

• Root Mean Square Error (RMSE):

• RMSE measures the average difference between corresponding pixels in the
original and reconstructed images.
• It is calculated as the square root of the average of squared differences between
corresponding pixel values.
• A lower RMSE value indicates better fidelity, as it means the reconstructed image
is closer to the original.
• RMS=√1/N*(i)−I(reconstructed)*(i)^2
 PSNR
• Peak Signal-to-Noise Ratio (PSNR):
• PSNR measures the ratio between the maximum possible power of a signal and the
power of corrupting noise that affects the fidelity of its representation.
• It is expressed in decibels (dB) and quantifies the quality of the reconstruction
relative to the maximum possible quality.
• A higher PSNR value indicates better fidelity, as it means the reconstructed image
has less distortion compared to the original.
• PSNR=10⋅log10​(MSEmax(Ioriginal​)2​)
• Where MSE is the mean squared error, calculated
MSE= 1/N−Ireconstructed​*(i)^2.
 SSIM
• Structural Similarity Index (SSIM):
• SSIM is a perceptual metric that compares local patterns of pixel intensities that have been
normalized for luminance and contrast.
• It measures the similarity between the original and reconstructed images in terms of luminance,
contrast, and structure.
• Higher SSIM values indicate better similarity between the original and reconstructed images.

SSIM(Ioriginal​,Ireconstructed​)=(μIoriginal​2​+μIreconstructed​2​+c1​)(σIoriginal​2​
+σIreconstructed​2​+c2​)(2μIoriginal​​μIreconstructed​​+c1​)(2σIoriginal​,Ireconstructed​​+c2​)​

• Where:
• μIoriginal​​,μIreconstructed​​are the means of the original and reconstructed images,
• σIoriginal​​,σIreconstructed​​are the standard deviations of the original and reconstructed images,
• σIoriginal​,Ireconstructed​​is the covariance of the original and reconstructed images,
• c1​and c2​are small constants to stabilize the division with weak denominator.
 RMS,PSNR,SSIM TABLE
CORRESPONDING TO EACH ALPHA
Alpha PSNR
VALUES
RMS SSIM
0 7.4131 108.6135 0.51813
0.1 8.1511 99.7665 0.51981
0.2 8.9491 91.009 0.56595
0.3 9.8126 82.3971 0.62036
0.4 10.7513 73.9562 0.67612
0.5 11.7733 65.7465 0.7288
0.6 12.8683 57.4598 0.77494
0.7 14.0463 50.6088 0.81381
0.8 15.2434 44.093 0.84431
0.9 16.375 38.7072 0.86706
1.0 17.236 35.0542 0.88247
1.1 17.630 33.4997 0.89126
1.2 17.4897 34.0451 0.89479
 RESULTS
• Result of preprocessing
Result of alpha blending
ADVANTAGES AND DISADVANTAGES
ADVANTAGES DISADVANTAGES

1. Simple Implementation 1. Dependence on Blending Coefficients

2. Control Over Transparency 2. Alignment of multiple images is crucial

3. Smooth Transitions 3. Artifacts and Visual Discontinuities

4. Compatibility 4. Limited Adaptability

5. Preservation of Information
 FUTURE SCOPE
• Real time medical image fusion for surgical navigation
Alpha blending mixes pictures together in surgery to help doctors see important information
from scans while they're operating. It can adjust how see-through these images are, making
it easier for doctors to navigate during surgery and make better decisions.

• promising avenue for combining CT and MRI scans

Alpha blending presents a promising avenue for combining CT and MRI scans. By blending
these images together, doctors can create a comprehensive view that incorporates the
strengths of both modalities.

• Interactive augmented reality(AR) for education and learning

• AI driven assistive technologies for image-guided interventions

 CONCLUSION
• This project introduces a novel approach to image restoration by combining image fusion
and alpha blending techniques. By leveraging information from multiple sources, our
method produces composite images with enhanced detail and clarity. Preprocessing
steps are first applied to prepare input images, followed by image fusion to integrate
their information effectively. Alpha blending is then used to refine the fused image,
providing precise control over each source image's contribution at the pixel level. This
fine-tuned control enhances the final image by emphasizing specific features, reducing
artifacts, and potentially resulting in visually compelling and application-specific results.

• Moving forward, our focus will be on exploring advanced image fusion algorithms and
developing new techniques for alpha channel generation. These refinements aim to
further improve the effectiveness and versatility of the image restoration process. By
continually innovating and refining our methodologies, we aim to advance the state-of-
the-art in image restoration, ultimately enhancing the quality and usefulness of medical
imaging and other fields that rely on high-fidelity visual data.
 REFERENCES
• Medical Imaging and Radiological Anatomy Dataset:
• Website: Kenhub. (n.d.). Medical Imaging and Radiological Anatomy. Retrieved from
https://www.kenhub.com/en/library/anatomy/medical-imaging-and-radiological-anatomy
• SIIM Medical Images Dataset:
• Kaggle. (n.d.). SIIM Medical Images. Retrieved from
https://www.kaggle.com/datasets/kmader/siim-medical-images
• Enhancement of Medical Images Using Image Processing in Matlab:
• Author(s): Unknown.
• Publication Title: Academia.edu.
• URL:
https://www.academia.edu/keypass/RXlHa3FGdDVZN01pdzRyQ0FvNGhiakV6VjdvQ2E2
MzFOcDhzVXBsckUxWTOtLXUvcFRQ79fb690f8b534360cc32becb33463ba1336503eO/
t/AqScyRJq308iwhwaE/resource/work/50382431/
Enhancement_Of_Medical_Images_Using_Image_Processing_In_Matlab?
email_work_card=reading-history
5. Medical Image Enhancement Application Using Histogram Equalization in Computational Libraries:
• Author(s): Unknown., Publication Title: Academia.edu.
• URL:https://www.academia.edu/keypass/
RXlHa3FGdDVZN01pdzRyQ0FvNGhiakV6VjdvQ2E2MzFOcDhzVXBsckUxWTOtLXUvcFRQ79fb
690f8b534360cc32becb33463ba1336503eO/t/AqScyRJq308iwhwaE/resource/work/95250335/
Paper1_Medical_Image_Enhancement_Application_Using_Histogram_Equalization_in_Computational
_Libraries?email_work_card=title
6. A Novel Approach for Contrast Enhancement and Noise Removal of Medical Images:
• Author(s): Unknown, Publication Title: Academia.edu.
• URL:https://www.academia.edu/keypass/
anU3cXpMd31HUmcwYVZwZUpsK2FkS011ZTB1a21GRzBuT3B4VOlLQkpTWTOtLTRuePBRHQ
pyTN-CXhB3/resource/work/78960132/
A_novel_approach_for_contrast_enhancement_and_noise_removal_of_medical_images?
email_work_card=title
7. A Novel Approach for Contrast Enhancement in Biomedical Images Based on Histogram
Equalization:
• Author(s): Unknown, Publication Title: Academia.edu.
• URL:[https://www.academia.edu/keypass/
aFVLbT15Q080d3FjUklmYUgzTmxmTlZ2SjV1by8xa05kSVpYTWNqcFNJazOtLTNFTk10TGZFSE
FHcVlrROpUcD15Ymc9PQ=d5e83a87d3ef53c0368aee4c3c4ec2a7ead23d5d/t/AqScyRH4EWUm-
08mZd/resource/worW356379/
A_Novel_Approach_for_Contrast_Enhancement_In_Biomedical_Images_Based_on_Histogram_Equal
THANK YOU