HTI 5720 Digital Imaging and PACS Experimental Tasks

Task 1: Study Title

State a title of your mini Digital Imaging Research Study.

I want to study the different filtering 'power' on the different filters applying on the image processing and aim to improve the image quality for clinical diagnosis purposes. In this project I am going to simulate the proposed solution on the assignment given in this course and show the difficulties by applying the simple High-pass, Low-pass and spatial filter therefore we do appraising the benefits and latest development of digital imaging and review the use of digital imaging technology in clinical practice. I wish the Canny Edge detector can reduce noise significantly with the semi-Low-dose CT images.

Task 2: Research Question and Objectives

State your research question, which has not been answered in the article reviewed in assignment 1, in one sentence.

The post-image processing can really help to reduce the unnecessary re-scanning processes when quality of images produced does not meet its best quality and the imperfect images still can give a sound evident-based radiological diagnosis.

Task 3: Experimental Design and Plan

Design your own research methods using image processing, analysis or visualization tools mentioned and taught in the practical sessions.

Write a brief plan within 250 words

There are huge amount of filter claims that they are good for image processing, and the Canny edge detector, according to the Wikipedia, claims to be the …

• good detection – the algorithm should mark as many real edges in the image as possible.
• good localization – edges marked should be as close as possible to the edge in the real image.
• minimal response – a given edge in the image should only be marked once, and where possible, image noise should not create false edges

(ref: http://en.wikipedia.org/wiki/Canny_edge_detector)

In this project I would like to test the canny Edge detector and conduct the in-depth study if it could apply to serve as an alternative noise filter on the low-dose CT- scanning proposed in the paper "Nonlinear Three-dimensional Noise Filter with Low-Dose CT Angiograpaphy"

First of all we use a good, clinically prefect image and treat it as a 'Gold Standard' for the prefect image. After that we add the noise into the images so that the image turns from prefect to 'imperfect'. We would like to apply different filter to test which is the best filter to show by Correlation function given from the MatLab.

Task 4: Results and Discussion

Report and discuss on the key findings obtained by implementing your plan and answer the research question you stated.

Narrate, illustrate and discuss on the key findings within 300 words and 2 figures/tables

1. Find the 'perfect image'
2. Apply the noise by the sharpening filter to simulate the Low-dose CT scanning and the noise found in the Low-dose CT images.
3. Apply the 'polluted' image into the MatLab
4. Apply the filter proposed

Original:

Filter applied with the Canny Edge Detector:

Original with noise applied:

Highpass filter applied:

Lowpass filter applied:

Sharpening filter applied:

(Pleas click here to check the fine graphics results)

Compare the Correlation value via MatLab

Filter applied	Canny Edge detector	Low Pass	High Pass	Sharpening
Result	0.9902	0.979	0.2031	0.3728
Preformance	Best	Good	Poor	Average

(1: Prefect match; 0: Totally un-match)

6. See the proposed idea can be achievable and while the image processing can give a sound clinical image for diagnosis proposed with low-dose CT scanning raised from the assignment paper.

Background of the Canny Edge detector:

Noise reduction

The Canny edge detector uses a filter based on the first derivative of a Gaussian, because it is susceptible to noise present on raw unprocessed image data, so to begin with, the raw image is convolved with a Gaussian filter. The result is a slightly blurred version of the original which is not affected by a single noisy pixel to any significant degree.

Here is an example of a 5x5 Gaussian filter, used to create the image to the right, with σ = 1.4:

Finding the intensity gradient of the image

A binary edge map, derived from the Sobel operator, with a threshold of 80. The edges are coloured to indicate the edge direction: yellow for 90 degrees, green for 45 degrees, blue for 0 degrees and red for 135 degrees.

An edge in an image may point in a variety of directions, so the Canny algorithm uses four filters to detect horizontal, vertical and diagonal edges in the blurred image. The edge detection operator (Roberts, Prewitt,Sobel for example) returns a value for the first derivative in the horizontal direction (Gy) and the vertical direction (Gx). From this the edge gradient and direction can be determined:

The edge direction angle is rounded to one of four angles representing vertical, horizontal and the two diagonals (0, 45, 90 and 135 degrees for example).

The Pseudo-code:

<First polltue the image by the sharpening-filter>

ct=dicomread('IM-0000.dcm');

w1 = [0 0 0;0 1 0;0 0 0];

w2 = ones(3,3)/9;

w3 = w1 - w2;

w4 = w1 - 0.9*w2;

ct_noise = imfilter(ct,w4,'conv');

<Canny algorithm>

B = [2 4 5 4 2;4 9 12 9 4;5 12 15 12 5;4 9 12 9 4;2 4 5 4 2];

wb = B/159;

ct_canny=imfilter(ct_noise,wb,'conv');

imtool(ct_canny, [0 200]);

<lowpass>

ct_lowpass=imfilter(ct_noise,w2,'conv');

imtool(ct_lowpass, [0 200]);

<highpass>

ct_highpass=imfilter(ct_noise,w3,'conv');

imtool(ct_highpass, [0 200]);

<sharpen>

ct_sharpen=imfilter(ct_noise,w4,'conv');

imtool(ct_sharpen, [0 200]);

<results>

RB=CORR2(ct_canny, ct)

RB=0.9902

R1=CORR2(ct_lowpass, ct)

R1=0.9790

R2=CORR2(ct_highpass, ct)

R2=0.2031

R3=CORR2(ct_sharpen, ct)

R3=0.3728