Asli Kumcu

Evaluating color difference measures in images

The most well known and widely used method for comparing two homogeneous color samples is the CIEDE2000 color difference formula because of its strong agreement with human perception. However, the formula is unreliable when applied over images and its spatial extensions have shown little improvement compared with the original formula. Hence, researchers have proposed many methods intending to measure color differences (CDs) in natural scene color images. However, these existing methods have not yet been rigorously compared. Therefore, in this work we review and evaluate CD measures with the purpose of answering the question to what extent do state-of-the-art CD measures agree with human perception of CDs in images? To answer the question, we have reviewed and evaluated eight state-of-the-art CD measures on a public image quality database. We found that the CIEDE2000, its spatial extension and the just noticeable CD measure perform well in computing CDs in images distorted by black level shift and color quantization algorithms (correlation higher than 0.8). However, none of the tested CD measures perform well on identifying CDs for the variety of color related distortions tested in this work, e.g., most of the tested CD measures showed a correlation lower than 0.65.

Visual quality assessment of H.264/AVC compressed laparoscopic video

The digital revolution has reached hospital operating rooms, giving rise to new opportunities such as tele-surgery and tele-collaboration. Applications such as minimally invasive and robotic surgery generate large video streams that demand gigabytes of storage and transmission capacity. While lossy data compression can offer large size reduction, high compression levels may significantly reduce image quality. In this study we assess the quality of compressed laparoscopic video using a subjective evaluation study and three objective measures. Test sequences were full High-Definition videos captures of four laparoscopic surgery procedures acquired on two camera types. Raw sequences were processed with H.264/AVC IPPP-CBR at four compression levels (19.5, 5.5, 2.8, and 1.8 Mbps). 16 non-experts and 9 laparoscopic surgeons evaluated the subjective quality and suitability for surgery (surgeons only) using Single Stimulus Continuous Quality Evaluation methodology. VQM, HDR-VDP-2, and PSNR objective measures were evaluated. The results suggest that laparoscopic video may be lossy compressed approximately 30 to 100 times (19.5 to 5.5 Mbps) without sacrificing perceived image quality, potentially enabling real-time streaming of surgical procedures even over wireless networks. Surgeons were sensitive to content but had large variances in quality scores, whereas non-experts judged all scenes similarly and over-estimated the quality of some sequences. There was high correlation between surgeons’ scores for quality and “suitability for surgery”. The objective measures had moderate to high correlation with subjective scores, especially when analyzed separately by camera type. Future studies should evaluate surgeons’ task performance to determine the clinical implications of conducting surgery with lossy compressed video.

Selecting stimuli parameters for video quality assessment studies based on perceptual similarity distances

This work presents a methodology to optimize the selection of multiple parameter levels of an image acquisition, degradation, or post-processing process applied to stimuli intended to be used in a subjective image or video quality assessment (QA) study. It is known that processing parameters (e.g. compression bit-rate) or technical quality measures (e.g. peak signal-to-noise ratio, PSNR) are often non-linearly related to human quality judgment, and the model of either relationship may not be known in advance. Using these approaches to select parameter levels may lead to an inaccurate estimate of the relationship between the parameter and subjective quality judgments – the system’s quality model. To overcome this, we propose a method for modeling the relationship between parameter levels and perceived quality distances using a paired comparison parameter selection procedure in which subjects judge the perceived similarity in quality. Our goal is to enable the selection of evenly sampled parameter levels within the considered quality range for use in a subjective QA study. This approach is tested on two applications: (1) selection of compression levels for laparoscopic surgery video QA study, and (2) selection of dose levels for an interventional X-ray QA study. Subjective scores, obtained from the follow-up single stimulus QA experiments conducted with expert subjects who evaluated the selected bit-rates and dose levels, were roughly equidistant in the perceptual quality space - as intended. These results suggest that a similarity judgment task can help select parameter values corresponding to desired subjective quality levels.

Influence of study design on digital pathology image quality evaluation : the need to define a clinical task

Abstract. Despite the current rapid advance in technologies for whole slide imaging, there is still no scientific consensus on the recommended methodology for image quality assessment of digital pathology slides. For medical images in general, it has been recommended to assess image quality in terms of doctors’ success rates in performing a specific clinical task while using the images (clinical image quality, cIQ). However, digital pathology is a new modality, and already identifying the appropriate task is difficult. In an alternative common approach, humans are asked to do a simpler task such as rating overall image quality (perceived image quality, pIQ), but that involves the risk of nonclinically relevant findings due to an unknown relationship between the pIQ and cIQ. In this study, we explored three different experimental protocols: (1) conducting a clinical task (detecting inclusion bodies), (2) rating image similarity and preference, and (3) rating the overall image quality. Additionally, within protocol 1, overall quality ratings were also collected (task-aware pIQ). The experiments were done by diagnostic veterinary pathologists in the context of evaluating the quality of hematoxylin and eosin-stained digital pathology slides of animal tissue samples under several common image alterations: additive noise, blurring, change in gamma, change in color saturation, and JPG compression. While the size of our experiments was small and prevents drawing strong conclusions, the results suggest the need to define a clinical task. Importantly, the pIQ data collected under protocols 2 and 3 did not always rank the image alterations the same as their cIQ from protocol 1, warning against using conventional pIQ to predict cIQ. At the same time, there was a correlation between the cIQ and task-aware pIQ ratings from protocol 1, suggesting that the clinical experiment context (set by specifying the clinical task) may affect human visual attention and bring focus to their criteria of image quality. Further research is needed to assess whether and for which purposes (e.g., preclinical testing) task-aware pIQ ratings could substitute cIQ for a given clinical task.

Effect of video lag on laparoscopic surgery: correlation between performance and usability at low latencies.

Few telesurgery studies assess the impact of latency on user experience, low latencies are often not studied despite evidence of negative effects, and some studies recruit inexperienced subjects instead of surgeons without evidence that latency affects both groups similarly.

How much image noise can be added in cardiac x-ray imaging without loss in perceived image quality?

Abstract. Cardiologists use x-ray image sequences of the moving heart acquired in real-time to diagnose and treat cardiac patients. The amount of radiation used is proportional to image quality; however, exposure to radiation is damaging to patients and personnel. The amount by which radiation dose can be reduced without compromising patient care was determined. For five patient image sequences, increments of computer-generated quantum noise (white + colored) were added to the images, frame by frame using pixel-to-pixel addition, to simulate corresponding increments of dose reduction. The noise adding software was calibrated for settings used in cardiac procedures, and validated using standard objective and subjective image quality measurements. The degraded images were viewed next to corresponding original (not degraded) images in a two-alternative-forced-choice staircase psychophysics experiment. Seven cardiologists and five radiographers selected their preferred image based on visualization of the coronary arteries. The point of subjective equality, i.e., level of degradation where the observer could not perceive a difference between the original and degraded images, was calculated; for all patients the median was 33%±15% dose reduction. This demonstrates that a 33%±15% increase in image noise is feasible without being perceived, indicating potential for 33%±15% dose reduction without compromising patient care.

Computing contrast ratio in images using local content information

It is well know that a measure of contrast in images is not yet fully defined. The conventional measures of contrast consist of global computations and therefore they have a poor performance. At the same time image quality assessment is often based on quantifying the visibility between a structure of interest or foreground and its surrounding background, i.e., the contrast ratio. Then, a high quality image is the one in which structures of interest are well distinguishable from the background. Therefore, the computation of contrast ratio is important in automatic image quality assessment and it should be computed locally taking into account the local distribution of pixel values. We estimate the contrast ratio by using Weber contrast in local image patches. The main contribution of this work lies in the characterization of local distribution of pixel values which is used for computing the contrast ratio. Here, local image patches are characterized by bimodal histograms representing a set of pixels which are likely to be inside the foreground and another set likely to be in the background. The local contrast ratio is estimated using the ratio between mean intensity values of each mode of the histogram. Our experimental results over two public image databases show that the proposed method is able to accurately predict changes of quality due to contrast decrements (Pearson correlations higher than 90%).

Estimating blur at the brain gray-white matter boundary for FCD detection in MRI

Focal cortical dysplasia (FCD) is a frequent cause of epilepsy and can be detected using brain magnetic resonance imaging (MRI). One important MRI feature of FCD lesions is the blurring of the gray-white matter boundary (GWB), previously modelled by the gradient strength. However, in the absence of additional FCD descriptors, current gradient-based methods may yield false positives. Moreover, they do not explicitly quantify the level of blur which prevents from using them directly in the process of automated FCD detection. To improve the detection of FCD lesions displaying blur, we develop a novel algorithm called iterating local searches on neighborhood (ILSN). The novelty is that it measures the width of the blurry region rather than the gradient strength. The performance of our method is compared with the gradient magnitude method using precision and recall measures. The experimental results, tested on MRI data of 8 real FCD patients, indicate that our method has higher ability to correctly identify the FCD blurring than the gradient method.

Volumetric detection tasks with varying complexity: Human observer performance

This study explores detection performance trends of human observers with respect to two parameters: task complexity determined by the frequency content of background and signal, and image viewing mode: singleslice (ss) versus multi-slice (ms) stack-browsing image presentation. The images are 3D correlated Gaussian noise with a 3D Gaussian signal centered in the image volume. To vary task complexity, we consider three different noise kernels while keeping the signal spread constant across all images. In ss mode, only the central slice of the volume is presented to the observer, while in ms mode all slices are available. All human readings are conducted in a controlled viewing environment on a 5MP digital mammography medical display. Overall, in line with the literature, we find that human performance increases in ms relative to ss image presentation mode. Furthermore, our experiments indicate that the extent of difference between ms and ss performance is influenced by the properties of image data (level of task complexity): the difference in performance increases (from ΔAUC= 0.14 to ΔAUC= 0.30) as the difference in the frequency content of the signal and the background increases. In other words, the benefit of having additional slices available in ms mode is larger for lower-complexity tasks. Future studies shall focus on comparing the results of the present study to the existing model observers for volumetric images, ultimately aiming to design an anthropomorphic model observer for volumetric detection tasks.

Validation of a new digital breast tomosynthesis medical display

The main objective of this study is to evaluate and validate the new Barco medical display MDMG-5221 which has been optimized for the Digital Breast Tomosynthesis (DBT) imaging modality system, and to prove the benefit of the new DBT display in terms of image quality and clinical performance. The clinical performance is evaluated by the detection of micro-calcifications inserted in reconstructed Digital Breast Tomosynthesis slices. The slices are shown in dynamic cine loops, at two frames rates. The statistical analysis chosen for this study is the Receiver Operating Characteristic Multiple-Reader, Multiple-Case methodology, in order to measure the clinical performance of the two displays. Four experienced radiologists are involved in this study. For this clinical study, 50 normal and 50 abnormal independent datasets were used. The result is that the new display outperforms the mammography display for a signal detection task using real DBT images viewed at 25 and 50 slices per second. In the case of 50 slices per second, the p-value = 0.0664. For a cut-off where alpha=0.05, the conclusion is that the null hypothesis cannot be rejected, however the trend is that the new display performs 6% better than the old display in terms of AUC. At 25 slices per second, the difference between the two displays is very apparent. The new display outperforms the mammography display by 10% in terms of AUC, with a good statistical significance of p=0.0415.