Algorithm for Morphological Cancer DetectionBackgroundGoalPowerPoint PresentationOriginal dataAfter Median filterAfter Unsharp MaskAfter ThresholdAfter FFTAfter LogAfter Mean FilterFinal graphConclusionsAlgorithm for Morphological Cancer DetectionCarmalyn LubawyMelissa SkalaECE 533 Fall 2004 ProjectBackground•Cancer diagnosis relies on morphological differences between normal and cancerous tissues •Cancerous cells are more disorganized than normal cells•Multiphoton microscopy has been used to evaluate cell morphology in normal and cancerous tissuesNormal Cancer2551280GoalDevelop an algorithm that provides relative differences in cell organization between multiphoton images of normal and cancerous tissues11. Fitzke, F.W., Fourier Transform Analysis of Human Corneal Endothelial Specular Photomicrographs. Exp Eye Res, 1997. 65Input ImageLogMedian FilterUnsharp MaskThresholdFFTMean FilterLine PlotAlgorithm FlowchartOriginal dataoriginal image Normal 1original image Normal 2original image Abnormal 1original image Abnormal 2Cancer 2Normal 1Normal 2Cancer 1After Median filterafter median filter Normal 1after median filter Normal 2after median filter Abnormal 1after median filter Abnormal 2Cancer 2Normal 1Normal 2Cancer 1After Unsharp Maskafter Gaussian filter Normal 1after Gaussian filter Normal 2after Gaussian filter Abnormal 1after Gaussian filter Abnormal 2Cancer 2Normal 1Normal 2Cancer 1After Thresholdafter threshold Normal 1after threshold Normal 2after threshold Abnormal 1after threshold Abnormal 2Cancer 2Normal 1Normal 2Cancer 1After FFTafter FFT Normal 1after FFT Normal 2after FFT Abnormal 1after FFT Abnormal 2Cancer 2Normal 1Normal 2Cancer 1After Logafter log Normal 1after log Normal 2after log Abnormal 1after log Abnormal 2Cancer 2Normal 1Normal 2Cancer 1After Mean Filterafter average filter Normal 1after average filter Normal 2after average filter Abnormal 1after average filter Abnormal 2Cancer 2Normal 1Normal 2Cancer 1Final graph-150 -100 -50 0 50 100 1500.550.60.650.70.750.80.850.90.951line plot Normal 1-150 -100 -50 0 50 100 1500.550.60.650.70.750.80.850.90.951line plot Normal 1-150 -100 -50 0 50 100 1500.60.650.70.750.80.850.90.951line plot Abnormal 2Cancer 2-150 -100 -50 0 50 100 1500.60.650.70.750.80.850.90.951line plot Abnormal 2Normal 1Normal 2Cancer 1Conclusions• Normal cells have peak at a spatial frequency of about 127 mm-1o Normal cells are about 7.8 m wide• Image processing allows for automatic evaluation of differences in cell organization between normal and cancerous
View Full Document
Unlocking...