Stephan M. Jonas

White-space models for offline Arabic handwriting recognition

We propose to explicitly model white-spaces for Arabic handwriting recognition within different writing variants. Position-dependent character shapes in Arabic handwriting allow for large white-spaces between characters even within words. Here, a separate character model for white-spaces in combination with a lexicon using different writing variants and character model length adaptation is proposed. Current handwriting recognition systems model the white-spaces implicitly within the character models leading to possibly degraded models, or try to explicitly segment the Arabic words into pieces of Arabic words being prone to segmentation errors. Several white-space modeling approaches are analyzed on the well known IFN/ENIT database and outperform the best reported error rates.

Microfluidic characterization of cilia-driven fluid flow using optical coherence tomography-based particle tracking velocimetry

Motile cilia are cellular organelles that generate directional fluid flow across various epithelial surfaces including the embryonic node and respiratory mucosa. The proper functioning of cilia is necessary for normal embryo development and, for the respiratory system, the clearance of mucus and potentially harmful particulate matter. Here we show that optical coherence tomography (OCT) is well-suited for quantitatively characterizing the microfluidic-scale flow generated by motile cilia. Our imaging focuses on the ciliated epithelium of Xenopus tropicalis embryos, a genetically manipulable and experimentally tractable animal model of human disease. We show qualitative flow profile characterization using OCT-based particle pathline imaging. We show quantitative, two-dimensional, two-component flow velocity field characterization using OCT-based particle tracking velocimetry. Quantitative imaging and phenotyping of cilia-driven fluid flow using OCT will enable more detailed research in ciliary biology and in respiratory medicine.

eMedOffice: A web-based collaborative serious game for teaching optimal design of a medical practice

BackgroundPreparing medical students for the takeover or the start-up of a medical practice is an important challenge in Germany today. Therefore, this paper presents a computer-aided serious game (eMedOffice) developed and currently in use at the RWTH Aachen University Medical School. The game is part of the attempt to teach medical students the organizational and conceptual basics of the medical practice of a general practitioner in a problem-based learning environment. This paper introduces methods and concepts used to develop the serious game and describes the results of an evaluation of the games application in curricular courses at the Medical School.ResultsResults of the conducted evaluation gave evidence of a positive learning effect of the serious game. Educational supervisors observed strong collaboration among the players inspired by the competitive gaming aspects. In addition, an increase in willingness to learn and the exploration of new self-invented ideas were observed and valuable proposals for further prospective enhancements were elicited. A statistical analysis of the results of an evaluation provided a clear indication of the positive learning effect of the game. A usability questionnaire survey revealed a very good overall score of 4.07 (5=best, 1=worst).ConclusionsWe consider web-based, collaborative serious games to be a promising means of improving medical education. The insights gained by the implementation of eMedOffice will promote the future development of more effective serious games for integration into curricular courses of the RWTH Aachen University Medical School.

Feature description with SIFT, SURF, BRIEF, BRISK, or FREAK? A general question answered for bone age assessment

Solving problems in medical image processing is either generic (being applicable to many problems) or specific (optimized for a certain task). For example, bone age assessment (BAA) on hand radiographs is a frequent but cumbersome task for radiologists. For this problem, many specific solutions have been proposed. However, general-purpose feature descriptors are used in many computer vision applications. Hence, the aim of this study is (i) to compare the five leading keypoint descriptors on BAA, and, in doing so, (ii) presenting a generic approach for a specific task. Two methods for keypoint selection were applied: sparse and dense feature points. For each type, SIFT, SURF, BRIEF, BRISK, and FREAK feature descriptors were extracted within the epiphyseal regions of interest (eROI). Classification was performed using a support vector machine. Reference data (1101 radiographs) of the University of Southern California was used for 5-fold cross-validation. The data was grouped into 30 classes representing the bone age range of 0-18 years. With a mean error of 0.605 years, dense SIFT gave best results and outperforms all published methods. The accuracy was 98.36% within the range of 2 years. Dense SIFT represents a generic method for a specific question.

Smartphone-based diagnostic for preeclampsia: an mHealth solution for administering the Congo Red Dot (CRD) test in settings with limited resources

OBJECTIVE Morbidity and mortality due to preeclampsia in settings with limited resources often results from delayed diagnosis. The Congo Red Dot (CRD) test, a simple modality to assess the presence of misfolded proteins in urine, shows promise as a diagnostic and prognostic tool for preeclampsia. We propose an innovative mobile health (mHealth) solution that enables the quantification of the CRD test as a batch laboratory test, with minimal cost and equipment. METHODS A smartphone application that guides the user through seven easy steps, and that can be used successfully by non-specialized personnel, was developed. After image acquisition, a robust analysis runs on a smartphone, quantifying the CRD test response without the need for an internet connection or additional hardware. In the first stage, the basic image processing algorithms and supporting test standardizations were developed using urine samples from 218 patients. In the second stage, the standardized procedure was evaluated on 328 urine specimens from 273 women. In the third stage, the application was tested for robustness using four different operators and 94 altered samples. RESULTS In the first stage, the image processing chain was set up with high correlation to manual analysis (z-test P < 0.001). In the second stage, a high agreement between manual and automated processing was calculated (Lins concordance coefficient ρc = 0.968). In the last stage, sources of error were identified and remedies were developed accordingly. Altered samples resulted in an acceptable concordance with the manual gold-standard (Lins ρc = 0.914). CONCLUSION Combining smartphone-based image analysis with molecular-specific disease features represents a cost-effective application of mHealth that has the potential to fill gaps in access to health care solutions that are critical to reducing adverse events in resource-poor settings.

Quantitative optical coherence tomography imaging of intermediate flow defect phenotypes in ciliary physiology and pathophysiology

Cilia-driven fluid flow is a critical yet poorly understood aspect of pulmonary physiology. Here, we demonstrate that optical coherence tomography-based particle tracking velocimetry can be used to quantify subtle variability in cilia-driven flow performance in Xenopus, an important animal model of ciliary biology. Changes in flow performance were quantified in the setting of normal development, as well as in response to three types of perturbations: mechanical (increased fluid viscosity), pharmacological (disrupted serotonin signaling), and genetic (diminished ciliary motor protein expression). Of note, we demonstrate decreased flow secondary to gene knockdown of kif3a, a protein involved in ciliogenesis, as well as a dose-response decrease in flow secondary to knockdown of dnah9, an important ciliary motor protein.

Analysis of Craniocardiac Malformations in Xenopus using Optical Coherence Tomography

Birth defects affect 3% of children in the United States. Among the birth defects, congenital heart disease and craniofacial malformations are major causes of mortality and morbidity. Unfortunately, the genetic mechanisms underlying craniocardiac malformations remain largely uncharacterized. To address this, human genomic studies are identifying sequence variations in patients, resulting in numerous candidate genes. However, the molecular mechanisms of pathogenesis for most candidate genes are unknown. Therefore, there is a need for functional analyses in rapid and efficient animal models of human disease. Here, we coupled the frog Xenopus tropicalis with Optical Coherence Tomography (OCT) to create a fast and efficient system for testing craniocardiac candidate genes. OCT can image cross-sections of microscopic structures in vivo at resolutions approaching histology. Here, we identify optimal OCT imaging planes to visualize and quantitate Xenopus heart and facial structures establishing normative data. Next we evaluate known human congenital heart diseases: cardiomyopathy and heterotaxy. Finally, we examine craniofacial defects by a known human teratogen, cyclopamine. We recapitulate human phenotypes readily and quantify the functional and structural defects. Using this approach, we can quickly test human craniocardiac candidate genes for phenocopy as a critical first step towards understanding disease mechanisms of the candidate genes.

Bone age assessment meets SIFT

Bone age assessment (BAA) is a method of determining the skeletal maturity and finding the growth disorder in the skeleton of a person. BAA is frequently used in pediatric medicine but also a time-consuming and cumbersome task for a radiologist. Conventionally, the Greulich and Pyle and the Tanner and Whitehouse methods are used for bone age assessment, which are based on visual comparison of left hand radiographs with a standard atlas. We present a novel approach for automated bone age assessment, combining scale invariant feature transform (SIFT) features and support vector machine (SVM) classification. In this approach, (i) data is grouped into 30 classes to represent the age range of 0- 18 years, (ii) 14 epiphyseal ROIs are extracted from left hand radiographs, (iii) multi-level image thresholding, using Otsu method, is applied to specify key points on bone and osseous tissues of eROIs, (iv) SIFT features are extracted for specified key points for each eROI of hand radiograph, and (v) classification is performed using a multi-class extension of SVM. A total of 1101 radiographs of University of Southern California are used in training and testing phases using 5- fold cross-validation. Evaluation is performed for two age ranges (0-18 years and 2-17 years) for comparison with previous work and the commercial product BoneXpert, respectively. Results were improved significantly, where the mean errors of 0.67 years and 0.68 years for the age ranges 0-18 years and 2-17 years, respectively, were obtained. Accuracy of 98.09 %, within the range of two years was achieved.

Wearable technology as a booster of clinical care

Wearable technology defines a new class of smart devices that are accessories or clothing equipped with computational power and sensors, like Google Glass. In this work, we propose a novel concept for supporting everyday clinical pathways with wearable technology. In contrast to most prior work, we are not focusing on the omnipresent screen to display patient information or images, but are trying to maintain existing workflows. To achieve this, our system supports clinical staff as a documenting observer, only intervening adequately if problems are detected. Using the example of medication preparation and administration, a task known to be prone to errors, we demonstrate the full potential of the new devices. Patient and medication identifier are captured with the built-in camera, and the information is send to a transaction server. The server communicates with the hospital information system to obtain patient records and medication information. The system then analyses the new medication for possible side-effects and interactions with already administered drugs. The result is sent to the device while encapsulating all sensitive information respecting data security and privacy. The user only sees a traffic light style encoded feedback to avoid distraction. The server can reduce documentation efforts and reports in real-time on possible problems during medication preparation or administration. In conclusion, we designed a secure system around three basic principles with many applications in everyday clinical work: (i) interaction and distraction is kept as low as possible; (ii) no patient data is displayed; and (iii) device is pure observer, not part of the workflow. By reducing errors and documentation burden, our approach has the capability to boost clinical care.

Tumor Volume as a Prognostic Factor in Resectable Malignant Melanoma

Background: Vertical tumor thickness according to Breslow and histological ulceration are still the most powerful predictors for the clinical outcome of resectable cutaneous malignant melanoma (MM) without lymph node infiltration. It has been proposed that tumor volume in MM may also be of prognostic relevance. Methods: We retrospectively analyzed the prognostic impact of tumor volume and other established risk factors in 122 MM patients with a median follow-up period of 39.7 months. Results: We found the logarithmic tumor volume to be a better prognostic factor compared to Breslow tumor thickness in multivariate analysis. MM with a tumor volume below a threshold of 140 mm3 had a significantly higher relapse-free survival after 5 years of 98% compared to 47% in larger MMs (p < 0.0001). Conclusion: In some melanomas with a low tumor thickness, a higher tumor volume appeared to be linked to a higher risk of disease recurrence. Inclusion of tumor volume into the risk assessment of resectable MM may be of benefit in the future.