In this paper we report the set-up and results of the Multimodal Brain Tumor Image Segmentation Benchmark (BRATS) organized in conjunction with the MICCAI 2012 and 2013 conferences Twenty state-of-the-art tumor segmentation algorithms were applied to a set of 65 multi-contrast MR scans of low- and high-grade glioma patients—manually annotated by up to four raters—and to 65 comparable scans generated using tumor image simulation software Quantitative evaluations revealed considerable disagreement between the human raters in segmenting various tumor sub-regions (Dice scores in the range 74%–85%), illustrating the difficulty of this task We found that different algorithms worked best for different sub-regions (reaching performance comparable to human inter-rater variability), but that no single algorithm ranked in the top for all sub-regions simultaneously Fusing several good algorithms using a hierarchical majority vote yielded segmentations that consistently ranked above all individual algorithms, indicating remaining opportunities for further methodological improvements The BRATS image data and manual annotations continue to be publicly available through an online evaluation system as an ongoing benchmarking resource

/pdf/the-multimodal-brain-tumor-image-segmentation-benchmark-2gfraj98dk.pdf

The Multimodal Brain Tumor Image Segmentation Benchmark (BRATS)

The Visual Display of Quantitative Information

With an estimated 160,000 deaths in 2018, lung cancer is the most common cause of cancer death in the United States1. Lung cancer screening using low-dose computed tomography has been shown to reduce mortality by 20–43% and is now included in US screening guidelines1–6. Existing challenges include inter-grader variability and high false-positive and false-negative rates7–10. We propose a deep learning algorithm that uses a patient’s current and prior computed tomography volumes to predict the risk of lung cancer. Our model achieves a state-of-the-art performance (94.4% area under the curve) on 6,716 National Lung Cancer Screening Trial cases, and performs similarly on an independent clinical validation set of 1,139 cases. We conducted two reader studies. When prior computed tomography imaging was not available, our model outperformed all six radiologists with absolute reductions of 11% in false positives and 5% in false negatives. Where prior computed tomography imaging was available, the model performance was on-par with the same radiologists. This creates an opportunity to optimize the screening process via computer assistance and automation. While the vast majority of patients remain unscreened, we show the potential for deep learning models to increase the accuracy, consistency and adoption of lung cancer screening worldwide. A convolutional neural network performs automated prediction of malignancy risk of pulmonary nodules in chest CT scan volumes and improves accuracy of lung cancer screening.

End-to-end lung cancer screening with three-dimensional deep learning on low-dose chest computed tomography

Image enhancement plays an important role in image processing and analysis. Among various enhancement algorithms, Retinex-based algorithms can efficiently enhance details and have been widely adopted. Since Retinex-based algorithms regard illumination removal as a default preference and fail to limit the range of reflectance, the naturalness of non-uniform illumination images cannot be effectively preserved. However, naturalness is essential for image enhancement to achieve pleasing perceptual quality. In order to preserve naturalness while enhancing details, we propose an enhancement algorithm for non-uniform illumination images. In general, this paper makes the following three major contributions. First, a lightness-order-error measure is proposed to access naturalness preservation objectively. Second, a bright-pass filter is proposed to decompose an image into reflectance and illumination, which, respectively, determine the details and the naturalness of the image. Third, we propose a bi-log transformation, which is utilized to map the illumination to make a balance between details and naturalness. Experimental results demonstrate that the proposed algorithm can not only enhance the details but also preserve the naturalness for non-uniform illumination images.

Naturalness Preserved Enhancement Algorithm for Non-Uniform Illumination Images

Validation, comparison, and combination of algorithms for automatic detection of pulmonary nodules in computed tomography images: The LUNA16 challenge.

Presence of emphysema is recognized to be one of the single most significant risk factors in risk models for the
prediction of lung cancer. Therefore, an automatically computed emphysema score would be a prime candidate as an
additional numerical feature for computer aided diagnosis (CADx) for indeterminate pulmonary nodules. We have
applied several histogram-based emphysema scores to 460 thoracic CT scans from the IDRI CT lung image database,
and analyzed the emphysema scores in conjunction with 3000 nodule malignancy ratings of 1232 pulmonary nodules
made by expert observers. Despite the emphysema being a known risk factor, we have not found any impact on the
readers' malignancy rating of nodules found in a patient with higher emphysema score. We have also not found any
correlation between the number of expert-detected nodules in a patient and his emphysema score, or the relative
craniocaudal location of the nodules and their malignancy rating. The inter-observer agreement of the expert ratings was
excellent on nodule diameter (as derived from manual delineations), good for calcification, and only modest for
malignancy and shape descriptions such as spiculation, lobulation, margin, etc.

Correlation of emphysema score with perceived malignancy of pulmonary nodules: a multi-observer study using the LIDC-IDRI CT lung database

In a medical screening workflow, a patient is disposed on a patient support (22) of a computed tomography imaging scanner (10). Baseline image data of the patient are acquired using the computed tomography imaging scanner. At least one medical diagnosis operation is performed while the patient remains on the patient support. Baseline image data are processed using a computer-aided lesion detection process performed as a background process during the at least one medical diagnosis operation. Conditional upon the processing detecting at least one lesion: (i) imaging parameters are determined including at least initial and final axial scanner positions for the at least one lesion based on results of the computer-aided lesion detection process; and (ii) an image of the at least one lesion is acquired with the patient remaining on the patient support using the computed tomography imaging scanner configured with the determined imaging parameters.

Cardiopulmonary Screening with Feedback From Analysis to Acquisition

The present invention relates to an image processing device for detecting line structures in an image data set. The device comprises a model definition unit (12) for defining a line model of a line structure to be detected, said line model comprising a number of voxels, a calculation unit (14) for calculating, per voxel of interest of said image data set, several correlation values of a correlation between said line model and an image area around said voxel of interest, said image area comprising a corresponding number of voxels as said line model, wherein for each of a number of different relative orientations of said line model with respect to said image area a respective correlation value is calculated, and a determining unit (16) for determining, per voxel of interest, the maximum correlation value from said calculated correlation values and the corresponding optimal orientation at which said maximum correlation value is obtained.

Image processing device and method

A method includes determining a change in a volume of a tissue of interest located in at least two data sets between the at least two data sets. The at least two image data sets include a first image data set acquired at a first time and a second image data set acquired at a second time, and the first and second times are different. The method includes generating a rendering which includes a region in which the tissue of interest is located and indicia that indicates a magnitude of the change across the region. The region is superimposed over the rendering, which is generated based on at least one of the at least two image data sets, and linked to a corresponding image respectively in the at least two image data sets including voxels representing tissue of interest. The method includes visually presenting the rendering in a graphical user interface.

Computer-aided identification of a tissue of interest

The present invention relates to a method for performing computer-aided detection (CAD) of a disease, e.g. lung tumours, on a medical image data set (20) from a imaging modality, such as MRI or CT. Initially, there is perform a segmentation of the medical image data set (20) using an anatomical model. Secondly, the segmented data is analyzed for characteristics of the disease resulting in a set of analysis data (25), and finally the set of analysis data (25) is evaluating with respect to the disease. At least one of these steps comprises as an input a position dependent probability (P_r) for the disease. The invention isadvantageous in that more efficient computations can be performed because the degree ofanalysis in a certain region of the part of the patient, e.g. the lung, can be adjusted or tailored to the level of probability of the disease in the that region. It is thereby possible to increase computational speed and thereby diseases like cancer, in particular cancer nodules in the lungs, can be more effectively found from medical image analysis.

Rafael Wiemker

Papers

Correlation of emphysema score with perceived malignancy of pulmonary nodules: a multi-observer study using the LIDC-IDRI CT lung database

Cardiopulmonary Screening with Feedback From Analysis to Acquisition

Image processing device and method

Computer-aided identification of a tissue of interest

Computer-aided detection (cad) of a disease