Trigger 5

finals

involves a series of techniques used to improve the quality and usability of radiographic images for analysis and diagnosis. These techniques aim to reduce noise, enhance contrast, and remove artifacts that may hinder accurate interpretation.

pre-medical image processing

refers to the process of converting raw data collected by an imaging system into a visual image that can be interpreted by a radiologist. This is especially important in advanced imaging techniques like computed tomography (CT), but it also applies in simpler forms to digital radiography.

image reconstruction

Types of Image Reconstruction:

• Analog to Digital Conversion (DR/CR)

CT Reconstruction (more complex):

• Filtered Back Projection (FBP)

• Iterative Reconstruction (IR)

• Fourier Transform Methods

Converts raw electrical signals from detectors into a grayscale digital image. Includes preprocessing and basic reconstruction of intensity patterns.

Analog to Digital Conversion (DR/CR)

A fast and commonly used algorithm that uses mathematical filters to reconstruct images from projections.

Filtered Back Projection (FBP):

A more advanced method that improves image quality and reduces noise or radiation dose.

Iterative Reconstruction (IR):

Uses frequency domain data to reconstruct images, sometimes used in MRI and CT.

Fourier Transform Methods

is a CT image reconstruction algorithm that combines filtering with back projection to create images from X-ray projections.

It involves two main steps: filtering the projection data to remove blurring and then back-projecting the filtered data to reconstruct the image.

Filtered Back Projection (FBP)

is the process of eliminating unwanted or non-diagnostic background signals from an X-ray image to improve clarity and focus on the relevant anatomical structures.

Background removal

Types of background removal

• Digital subtraction

• Histogram Equalization or Thresholding

• Edge Detection and Cropping

• Flat-Field Correction

refers to Digital Subtraction Angiography (DSA). It's a fluoroscopic technique used to visualize blood vessels by digitally subtracting non-vessel structures like bones and soft tissues from the images, leaving only the blood vessels visible.

Digital subtraction

are image processing techniques used for enhancing contrast and simplifying images.

Histogram equalization and thresholding

redistributes pixel values to spread them across the entire intensity range,

histogram equalization

converts grayscale images into binary images by setting a specific intensity value as a boundary.

thresholding

helps identify boundaries of objects in medical images and removes unnecessary portions of the image.

Edge Detection and Cropping

a technique used to reduce image artifacts caused by non-uniformities in the X-ray beam or the detector itself. These artifacts can manifest as shading, inconsistencies in pixel response, or variations in intensity across the image.

Flat-Field Correction

is the process of removing or reducing the noise from the image.

Noise removal algorithm

refers to unwanted variations in an image that don't represent the subject's anatomy.

noise

What causes noise?

• Quantum mottle

• Electronic Noise

• Structured noise

This is the most prevalent type of noise in X-ray imaging, including plain film radiography, mammography, and CT. It arises from the random nature of X-ray photon interactions with the detector. The more X-rays used to create the image, the less quantum mottle there will be.

quantum mottle

can arise from various sources, including the detector itself and the analog-to-digital converter.

Electronic Noise

unwanted, repeating patterns or artifacts in an X-ray image that resemble anatomical structures or consistent patterns, making it harder to distinguish real anatomy from false features.

Structured noise

What are the types of noises?

• Salt and pepper noise

• Speckle noise

• Poisson noise

• Gaussian noise

also known as impulse noise, is a type of image degradation characterized by sparsely occurring black and white pixels, appearing as scattered dots on an image.

Salt-and-pepper noise,

often associated with ultrasound, is a granular, textured appearance in the image due to the interference of echoes from multiple scattering points.

Speckle noise

a type of random noise in digital radiography, arises from the statistical nature of X-ray photon detection. It's particularly prevalent in low-dose imaging,

Poisson noise

a type of random noise following a normal distribution, can impact the quality of radiographic images, especially in digital radiography and other imaging modalities like CT and MRI.

Gaussian noise

is the process of reducing the size of digital radiographic image files while maintaining as much diagnostic image quality as possible.

Image compression

2 types of image compression:

• Lossless Compression

• Lossy Compression

Type of image compression that

• Preserves all original data

• Every bit of information in the original image is retained after compression and decompression.

• No quality degradation

• The compressed image is identical to the original, ensuring a perfect copy.

Lossless Compression

Type of image compression that

• Reduces file size by removing some data

• Less critical or redundant information is discarded to achieve smaller file sizes.

• Potential for quality loss

• The compressed image may appear slightly different from the original, especially at high compression levels.

Lossy Compression

Common preprocessing steps include

• denoising

• contrast enhancement,

• background removal

• image registration.