Automated classification of duodenal imagery in celiac disease using evolved Fourier feature vectors

Abstract

Feature extraction techniques based on selection of highly discriminant Fourier filters have been developed for an automated classification of magnifying endoscope images with respect to pit patterns of colon lesions. These are applied to duodenal imagery for diagnosis of celiac disease. Features are extracted from the Fourier domain by selecting the most discriminant features using an evolutionary algorithm. Subsequent classification is performed with various standard algorithms (KNN, SVM, Bayes classifier) and combination of several Fourier filters and classifiers which is called multiclassifier.The obtained results are promising, due to a high specificity for the detection of mucosal damage typical of untreated celiac disease.

Quantitative Assessment of Endoscopic Images for Degree of Villous Atrophy in Celiac Disease

“…Quantitative image processing applied to capsule endoscopy can potentially be useful to assess the patchiness of disease [23][24][25][26] and improvement in mucosal appearance after institution of a gluten-free diet [10], a possible future application of our method. To validate the usefulness of our new method in cases of patchy lesions will require acquisition of endoscopic biopsies at each subimage location where L is measured.…”

Automated Marsh-like classification of celiac disease in children using local texture operators

“…Previous experiments with the employed feature extraction techniques have shown that these results can be further improved by employing SVM or Bayes classifiers (Hegenbart et al, 2009;Vécsei et al, 2009).…”

“…To improve the results obtained by the k-nearest neighbor classifier, we use an Ensemble classifier as described in Häfner et al (2009b); Vécsei et al (2009). This classifier aims at achieving a higher overall classification accuracy and more stable results across different image classes by combining different methods.…”

Improved endoscope distortion correction does not necessarily enhance mucosa-classification based medical decision support systems

“…An automated system identifying areas affected by celiac disease in the duodenum can help to improve biopsy reliability (by indicating areas eventually affected by celiac disease), can aid to improve less invasive diagnosis techniques avoiding biopsies, and can reduce the costs of interpreting video material captured during capsule endoscopy [13]. Prior approaches dealing with the computer-aided diagnosis of celiac disease using endoscopic still images images include feature extraction based on Local Binary Pattern based operators [13], band-pass type Fourier filters [14], histogram and wavelet-transform based features as well as smoothness/sharpness measures [13]. Impact of different image capturing techniques and various pre-processing schemes has been investigated as well [15].…”