Tsubasa Hirakawa

SVM-MRF segmentation of colorectal NBI endoscopic images.

In this paper we investigate a method for segmentation of colorectal Narrow Band Imaging (NBI) endoscopic images with Support Vector Machine (SVM) and Markov Random Field (MRF). SVM classifiers recognize each square patch of an NBI image and output posterior probabilities that represent how likely the given patch falls into a certain label. To prevent the spatial inconsistency between adjacent patches and encourage segmented regions to have smoother shapes, MRF is introduced by using the posterior outputs of SVMs as a unary term of MRF energy function. Segmentation results of 1191 NBI images are evaluated in experiments in which SVMs were trained with 480 trimmed NBI images and the MRF energy was minimized by an α - β swap Graph Cut.

Transfer learning for Bag-of-Visual words approach to NBI endoscopic image classification

We address a problem of endoscopic image classification taken by different (e.g., old and new) endoscopies. Our proposed method formulates the problem as a constraint optimization that estimates a linear transformation between feature vectors (or Bag-of-Visual words histograms) in a framework of transfer learning. Experimental results show that the proposed method works much better than the case without feature transformation.

Labeling colorectal NBI zoom-videoendoscope image sequences with MRF and SVM

In this paper, we propose a sequence labeling method by using SVM posterior probabilities with a Markov Random Field (MRF) model for colorectal Narrow Band Imaging (NBI) zoom-videoendoscope. Classifying each frame of a video sequence by SVM classifiers independently leads to an output sequence which is unstable and hard to understand by endoscopists. To make it more stable and readable, we use an MRF model to label the sequence of posterior probabilities. In addition, we introduce class asymmetry for the NBI images in order to keep and enhance frames where there is a possibility that cancers might have been detected. Experimental results with NBI video sequences demonstrate that the proposed MRF model with class asymmetry performs much better than a model without asymmetry.

Discriminative Subtree Selection for NBI Endoscopic Image Labeling

In this paper, we propose a novel method for image labeling of colorectal Narrow Band Imaging (NBI) endoscopic images based on a tree of shapes. Labeling results could be obtained by simply classifying histogram features of all nodes in a tree of shapes, however, satisfactory results are difficult to obtain because histogram features of small nodes are not enough discriminative. To obtain discriminative subtrees, we propose a method that optimally selects discriminative subtrees. We model an objective function that includes the parameters of a classifier and a threshold to select subtrees. Then labeling is done by mapping the classification results of nodes of the subtrees to those corresponding image regions. Experimental results on a dataset of 63 NBI endoscopic images show that the proposed method performs qualitatively and quantitatively much better than existing methods.

Smoothing posterior probabilities with a particle filter of dirichlet distribution for stabilizing colorectal NBI endoscopy recognition

This paper proposes a method for smoothing the posterior probabilities obtained from classification results of time series input. We deal with this problem as a filtering problem with Dirichlet distribution and develop a particle filtering for this task. As a practical example of smoothing, we apply the proposed method to stabilizing NBI endoscopy recognition results over time. Experimental results demonstrate that our approach can effectively smooth highly unstable probability curves.

Tree-Wise Discriminative Subtree Selection for Texture Image Labeling

In this paper, we propose a method for texture image labeling that works with a small number of training images. Our method is based on a tree of shapes and histogram features computed on the tree structure. Labeling results could be obtained by simply classifying the histogram features of all nodes in a tree of shapes. However, it is difficult to obtain satisfactory results, because features of smaller nodes are not sufficiently discriminative. Consequently, our method selects optimal discriminative subtrees for image labeling. We model an objective function that includes the parameters of a classifier and a set of thresholds for each training image to be used to select optimal subtrees. Then, labeling is performed by mapping the classification results of selected subtrees into corresponding blobs in the image. Experimental results with synthetic and real data sets that we created for evaluation show that the proposed method performs qualitatively and quantitatively much better than the existing methods.

Defocus-aware Dirichlet particle filter for stable endoscopic video frame recognition.

BACKGROUND AND OBJECTIVE A computer-aided system for colorectal endoscopy could provide endoscopists with important helpful diagnostic support during examinations. A straightforward means of providing an objective diagnosis in real time might be for using classifiers to identify individual parts of every endoscopic video frame, but the results could be highly unstable due to out-of-focus frames. To address this problem, we propose a defocus-aware Dirichlet particle filter (D-DPF) that combines a particle filter with a Dirichlet distribution and defocus information. METHODS We develop a particle filter with a Dirichlet distribution that represents the state transition and likelihood of each video frame. We also incorporate additional defocus information by using isolated pixel ratios to sample from a Rayleigh distribution. RESULTS We tested the performance of the proposed method using synthetic and real endoscopic videos with a frame-wise classifier trained on 1671 images of colorectal endoscopy. Two synthetic videos comprising 600 frames were used for comparisons with a Kalman filter and D-DPF without defocus information, and D-DPF was shown to be more robust against the instability of frame-wise classification results. Computation time was approximately 88ms/frame, which is sufficient for real-time applications. We applied our method to 33 endoscopic videos and showed that the proposed method can effectively smoothen highly unstable probability curves under actual defocus of the endoscopic videos. CONCLUSION The proposed D-DPF is a useful tool for smoothing unstable results of frame-wise classification of endoscopic videos to support real-time diagnosis during endoscopic examinations.