Student : Yohann Sulaiman

Address: <DELETED>

Course:   B.Eng. (hons) Electrical & Electronic Engineering.

Project Supervisor: Dr. M.H.Fisher.

Title: MultiSlice RTP Environment

Student activities and output:

[1] Background Reading and introduction to C under X-Windows with
      MOTIF extensions
[2] Familiarisation with programming in X-Windows/MOTIF
[3] Design of a extension to the existing system
[4] Rewriting the existing system to work with multiple images
[5] Removing the old system and rewriting some of the old algorithms to
     work in the new system
[6] Testing/Debugging and improving the new code
[7] Inserting additional comments and improving the code to support extensions to it in the future
[8] Doing a formal report

Timetable :

Sept-Oct : Familiarisation with X-Windows/MOTIF under IRIX
Oct-Nov : Designing extensions/rewriting the old system
Nov-Dec: Final project specifications/Implementation of the new system
Dec-Jan : Rewriting all the algorithms mentioned in the formal spec. & removing the
                old system completely
Jan-Feb : Testing & Debugging phase. Improving the code to support extensions
Feb-Mar : Adding additional comments and writing explanatory files for future use.
Mar-Apr: Formal Report Completion / Bench assessments

Students Signature :                                            Supervisors Signature:

Date :                                                                      Date:

Attached: Formal Project Specifications.

Computer Aided Segmentation of Medical Tomographic Images - Clinical Evaluation System


Over the past 3 years we have developed a prototype system which has been a test-bed for a number of image segmentation algorithms. We now wish to further develop a clinical evaluation system which could be used by radiologists and consultants at Walsgrave Hospital. This will involve the re-design of the user interface.

The main difference between the clinical evaluation system and the existing system will be :-

Algorithms currently operating on single image `slices' only - the new system will operate on sequences of slices comprising `patient studies'. Typically a patient study will consist of 10 individual image `slices'.

There are currently many algorithms implemented in the experimental system. The clinical evaluation system will only need `user' segmentation algorithms.


We envisage the following functions :-

File Management:

        Load -  This operation will load a patient `study' (10 images/study) and display the
                      images as `thumbnails'. Each slice should be identifiable uniquely (usually by
                     the filename on the slice). In addition, load should have the capability to load
                     a single `slice' if required to do so.

        Save -  This operation allows the processed image to be saved to disk under a new

        Print - Print out a selected `slice' in a universal postscript format file to send to a
                    postscript printer. The slice should be rotated by 90 degrees automatically
                    if it cannot fit onto an A4 sized page.
        Maximise/Minimize - Maximize images or minimize them if required.
        Resize all - Resize all current images by maximising or minimising as reqd.
        Deletion - `Remove' a single image from the screen.
        Reorder - Some form of Tiling or Cascading images into an orderly form should be
                          available to the user.
        Editing Tools -    The user should be able to select a specified area (rectangular,
                                    circular etc.) and the selection should not affect the image in any
                                    way (i.e. it can be erased after selection).
        Zoom - Areas of full-size images can be magnified or zoomed.

Region Differential - `Compare' two images and test for similarities.
Region Growing:-
                Grow Region - This operation should be included and modified to work in 3D.
                Grow Image - This operation should be included and modified to work in 3D.
Watershed Transform:-
                Watershed - Perform a Gaussian filter followed by watershed segmentation
                on the image.
                Erode - Image regions can be eroded by a user defined amount.
                Dilate - Image regions can be dilated by a user defined amount.

We would also like a simple user interface with a `floating' menu bar and separate windows for each image. In addition, a clickable `button-selectable' method for selecting images would be useful.

     Dr. M.H. Fisher                                                                            Yohann Sulaiman

_________________                                                                     _______________
Supervisor's signature                                                                      Student's signature

Completed and verified by supervisor as of 4/12/97. Project report and documentation
now under construction.