Dataset Information

Training a cell-level classifier for detecting basal-cell carcinoma by combining human visual attention maps with low-level handcrafted features.

ABSTRACT: Computational histomorphometric approaches typically use low-level image features for building machine learning classifiers. However, these approaches usually ignore high-level expert knowledge. A computational model (M_im) combines low-, mid-, and high-level image information to predict the likelihood of cancer in whole slide images. Handcrafted low- and mid-level features are computed from area, color, and spatial nuclei distributions. High-level information is implicitly captured from the recorded navigations of pathologists while exploring whole slide images during diagnostic tasks. This model was validated by predicting the presence of cancer in a set of unseen fields of view. The available database was composed of 24 cases of basal-cell carcinoma, from which 17 served to estimate the model parameters and the remaining 7 comprised the evaluation set. A total of 274 fields of view of size [Formula: see text] were extracted from the evaluation set. Then 176 patches from this set were used to train a support vector machine classifier to predict the presence of cancer on a patch-by-patch basis while the remaining 98 image patches were used for independent testing, ensuring that the training and test sets do not comprise patches from the same patient. A baseline model (M_ex) estimated the cancer likelihood for each of the image patches. M_ex uses the same visual features as M_im, but its weights are estimated from nuclei manually labeled as cancerous or noncancerous by a pathologist. M_im achieved an accuracy of 74.49% and an [Formula: see text]-measure of 80.31%, while M_ex yielded corresponding accuracy and F-measures of 73.47% and 77.97%, respectively.

SUBMITTER: Corredor G

PROVIDER: S-EPMC5363808 | biostudies-other | 2017 Apr

REPOSITORIES: biostudies-other

ACCESS DATA

Publications

Training a cell-level classifier for detecting basal-cell carcinoma by combining human visual attention maps with low-level handcrafted features.

Corredor Germán G Whitney Jon J Arias Viviana V Madabhushi Anant A Romero Eduardo E

Journal of medical imaging (Bellingham, Wash.) 20170311 2

Computational histomorphometric approaches typically use low-level image features for building machine learning classifiers. However, these approaches usually ignore high-level expert knowledge. A computational model (M_im) combines low-, mid-, and high-level image information to predict the likelihood of cancer in whole slide images. Handcrafted low- and mid-level features are computed from area, color, and spatial nuclei distributions. High-level information is implicitly captured from the rec ...[more]

PMID: 28382314

Similar Datasets

Project description:Compassion meditation training is hypothesized to increase the motivational salience of cues of suffering, while also enhancing equanimous attention and decreasing emotional reactivity to suffering. However, it is currently unknown how compassion meditation impacts visual attention to suffering, and how this impacts neural activation in regions associated with motivational salience as well as aversive responses, such as the amygdala. Healthy adults were randomized to 2 weeks of compassion or reappraisal training. We measured BOLD fMRI responses before and after training while participants actively engaged in their assigned training to images depicting human suffering or non-suffering. Eye-tracking data were recorded concurrently, and we computed looking time for socially and emotionally evocative areas of the images, and calculated visual preference for suffering vs. non-suffering. Increases in visual preference for suffering due to compassion training were associated with decreases in the amygdala, a brain region involved in negative valence, arousal, and physiological responses typical of fear and anxiety states. This pattern was specifically in the compassion group, and was not found in the reappraisal group. In addition, compassion training-related increases in visual preference for suffering were also associated with decreases in regions sensitive to valence and empathic distress, spanning the anterior insula and orbitofrontal cortex (while the reappraisal group showed the opposite effect). Examining visual attention alone demonstrated that engaging in compassion in general (across both time points) resulted in visual attention preference for suffering compared to engaging in reappraisal. Collectively, these findings suggest that compassion meditation may cultivate visual preference for suffering while attenuating neural responses in regions typically associated with aversive processing of negative stimuli, which may cultivate a more equanimous and nonreactive form of attention to stimuli of suffering.

Dataset Information

Training a cell-level classifier for detecting basal-cell carcinoma by combining human visual attention maps with low-level handcrafted features.

Publications

Training a cell-level classifier for detecting basal-cell carcinoma by combining human visual attention maps with low-level handcrafted features.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets