A Food Recognition System for Diabetic Patients based on an Optimized Bag of Features Model

Computer vision-based food recognition could be used to estimate a meal's carbohydrate content for diabetic patients. This study proposes a methodology for automatic food recognition, based on the Bag of Features (BoF) model. An extensive technical investigation was conducted for the identification and optimization of the best performing components involved in the BoF architecture, as well as the estimation of the corresponding parameters. For the design and evaluation of the prototype system, a visual dataset with nearly 5,000 food images was created and organized into 11 classes. The optimized system computes dense local features, using the scale-invariant feature transform on the HSV color space, builds a visual dictionary of 10,000 visual words by using the hierarchical k-means clustering and finally classifies the food images with a linear support vector machine classifier. The system achieved classification accuracy of the order of 78%, thus proving the feasibility of the proposed approach in a very challenging image dataset.

Calibration of a surgical microscope with automated zoom lenses using an active optical tracker

BACKGROUND: In this paper, we present a new method for the calibration of a microscope and its registration using an active optical tracker. METHODS: Practically, both operations are done simultaneously by moving an active optical marker within the field of view of the two devices. The IR LEDs composing the marker are first segmented from the microscope images. By knowing their corresponding three-dimensional (3D) position in the optical tracker reference system, it is possible to find the transformation matrix between the referential of the two devices. Registration and calibration parameters can be extracted directly from that transformation. In addition, since the zoom and focus can be modified by the surgeon during the operation, we propose a spline based method to update the camera model to the new setup. RESULTS: The proposed technique is currently being used in an augmented reality system for image-guided surgery in the fields of ear, nose and throat (ENT) and craniomaxillofacial surgeries. CONCLUSIONS: The results have proved to be accurate and the technique is a fast, dynamic and reliable way to calibrate and register the two devices in an OR environment.

Retinal Laser Lesion Visibility in Simultaneous Ultra-High Axial Resolution Optical Coherence Tomography

Ex vivo porcine retina laser lesions applied with varying laser power (20 mW–2 W, 10 ms pulse, 196 lesions) are manually evaluated by microscopic and optical coherence tomography (OCT) visibility, as well as in histological sections immediately after the deposition of the laser energy. An optical coherence tomography system with 1.78 um axial resolution specifically developed to image thin retinal layers simultaneously to laser therapy is presented, and visibility thresholds of the laser lesions in OCT data and fundus imaging are compared. Optical coherence tomography scans are compared with histological sections to estimate the resolving power for small optical changes in the retinal layers, and real-time time-lapse scans during laser application are shown and analyzed quantitatively. Ultrahigh-resolution OCT inspection features a lesion visibility threshold 40–50 mW (17 reduction) lower than for visual inspection. With the new measurement system, 42 of the lesions that were invisible using state-of-the-art ophthalmoscopic methods could be detected.

Assessment of ultra-high resolution optical coherence tomography for monitoring tissue effects caused by laser photocoagulation of ex-vivo porcine retina

Retinal laser photocoagulation is an established and successful treatment for a variety of retinal diseases. While being a valuable treatment modality, laser photocoagulation shows the drawback of employing high energy lasers which are capable of physically destroying the neural retina. For reliable therapy, it is therefore crucial to closely monitor the therapy effects caused in the retinal tissue. A depth resolved representation of optical tissue properties as provided by optical coherence tomography may provide valuable information about the treatment effects in the retinal layers if recorded simultaneously to laser coagulation. Therefore, in this work, the use of ultra-high resolution optical coherence tomography to represent tissue changes caused by conventional and selective retinal photocoagulation is investigated. Laser lesions were placed on porcine retina ex-vivo using a 577 nm laser as well as a pulsed laser at 527 nm built for selective treatment of the retinal pigment epithelium. Applied energies were varied to generate lesions best representing the span from under- to overtreatment. The lesions were examined using a custom-designed optical coherence tomography system with an axial resolution of 1.78 μm and 70 kHz Ascan rate. Optical coherence tomography scans included volume scans before and after irradiation, as well as time lapse scans (Mscan) of the lesions. Results show OCT lesion visibility thresholds to be below the thresholds of ophthalmoscopic inspection. With the ultra-high resolution OCT, 42% - 44% of ophthalmoscopically invisible lesions could be detected and lesions that were under- or overexposed could be distinguished using the OCT data.

Evaluation of Three State-of-the-Art Classifiers for Recognition of Activities of Daily Living from Smart Home Ambient Data

Smart homes for the aging population have recently started attracting the attention of the research community. The "health state" of smart homes is comprised of many different levels; starting with the physical health of citizens, it also includes longer-term health norms and outcomes, as well as the arena of positive behavior changes. One of the problems of interest is to monitor the activities of daily living (ADL) of the elderly, aiming at their protection and well-being. For this purpose, we installed passive infrared (PIR) sensors to detect motion in a specific area inside a smart apartment and used them to collect a set of ADL. In a novel approach, we describe a technology that allows the ground truth collected in one smart home to train activity recognition systems for other smart homes. We asked the users to label all instances of all ADL only once and subsequently applied data mining techniques to cluster in-home sensor firings. Each cluster would therefore represent the instances of the same activity. Once the clusters were associated to their corresponding activities, our system was able to recognize future activities. To improve the activity recognition accuracy, our system preprocessed raw sensor data by identifying overlapping activities. To evaluate the recognition performance from a 200-day dataset, we implemented three different active learning classification algorithms and compared their performance: naive Bayesian (NB), support vector machine (SVM) and random forest (RF). Based on our results, the RF classifier recognized activities with an average specificity of 96.53%, a sensitivity of 68.49%, a precision of 74.41% and an F-measure of 71.33%, outperforming both the NB and SVM classifiers. Further clustering markedly improved the results of the RF classifier. An activity recognition system based on PIR sensors in conjunction with a clustering classification approach was able to detect ADL from datasets collected from different homes. Thus, our PIR-based smart home technology could improve care and provide valuable information to better understand the functioning of our societies, as well as to inform both individual and collective action in a smart city scenario.

Accident or homicide - virtual crime scene reconstruction using 3D methods

The analysis and reconstruction of forensically relevant events, such as traffic accidents, criminal assaults and homicides are based on external and internal morphological findings of the injured or deceased person. For this approach high-tech methods are gaining increasing importance in forensic investigations. The non-contact optical 3D digitising system GOM ATOS is applied as a suitable tool for whole body surface and wound documentation and analysis in order to identify injury-causing instruments and to reconstruct the course of event. In addition to the surface documentation, cross-sectional imaging methods deliver medical internal findings of the body. These 3D data are fused into a whole body model of the deceased. Additional to the findings of the bodies, the injury inflicting instruments and incident scene is documented in 3D. The 3D data of the incident scene, generated by 3D laser scanning and photogrammetry, is also included into the reconstruction. Two cases illustrate the methods. In the fist case a man was shot in his bedroom and the main question was, if the offender shot the man intentionally or accidentally, as he declared. In the second case a woman was hit by a car, driving backwards into a garage. It was unclear if the driver drove backwards once or twice, which would indicate that he willingly injured and killed the woman. With this work, we demonstrate how 3D documentation, data merging and animation enable to answer reconstructive questions regarding the dynamic development of patterned injuries, and how this leads to a real data based reconstruction of the course of event.

Comparison of freehand-navigated and aiming device-navigated targeting of liver lesions

BACKGROUND Accurate needle placement is crucial for the success of percutaneous radiological needle interventions. We compared three guiding methods using an optical-based navigation system: freehand, using a stereotactic aiming device and active depth control, and using a stereotactic aiming device and passive depth control. METHODS For each method, 25 punctures were performed on a non-rigid phantom. Five 1 mm metal screws were used as targets. Time requirements were recorded, and target positioning errors (TPE) were measured on control scans as the distance between needle tip and target. RESULTS Time requirements were reduced using the aiming device and passive depth control. The Euclidian TPE was similar for each method (4.6 ± 1.2-4.9 ± 1.7 mm). However, the lateral component was significantly lower when an aiming device was used (2.3 ± 1.3-2.8 ± 1.6 mm with an aiming device vs 4.2 ± 2.0 mm without). DISCUSSION Using an aiming device may increase the lateral accuracy of navigated needle insertion.

Incidence and Imaging Outcomes of Acute Scaffold Disruption and Late Structural Discontinuity After Implantation of the Absorb Everolimus-Eluting Fully Bioresorbable Vascular Scaffold: Optical Coherence Tomography Assessment in the ABSORB Cohort B Trial (A Clinical Evaluation of the Bioabsorbable Everolimus Eluting Coronary Stent System in the Treatment of Patients With De Novo Native Coronary Artery Lesions).

OBJECTIVES This study sought to describe the frequency and clinical impact of acute scaffold disruption and late strut discontinuity of the second-generation Absorb bioresorbable polymeric vascular scaffolds (Absorb BVS, Abbott Vascular, Santa Clara, California) in the ABSORB (A Clinical Evaluation of the Bioabsorbable Everolimus Eluting Coronary Stent System in the Treatment of Patients With De Novo Native Coronary Artery Lesions) cohort B study by optical coherence tomography (OCT) post-procedure and at 6, 12, 24, and 36 months. BACKGROUND Fully bioresorbable scaffolds are a novel approach to treatment for coronary narrowing that provides transient vessel support with drug delivery capability without the long-term limitations of metallic drug-eluting stents. However, a potential drawback of the bioresorbable scaffold is the potential for disruption of the strut network when overexpanded. Conversely, the structural discontinuity of the polymeric struts at a late stage is a biologically programmed fate of the scaffold during the course of bioresorption. METHODS The ABSORB cohort B trial is a multicenter single-arm trial assessing the safety and performance of the Absorb BVS in the treatment of 101 patients with de novo native coronary artery lesions. The current analysis included 51 patients with 143 OCT pullbacks who underwent OCT at baseline and follow-up. The presence of acute disruption or late discontinuities was diagnosed by the presence on OCT of stacked, overhung struts or isolated intraluminal struts disconnected from the expected circularity of the device. RESULTS Of 51 patients with OCT imaging post-procedure, acute scaffold disruption was observed in 2 patients (3.9%), which could be related to overexpansion of the scaffold at the time of implantation. One patient had a target lesion revascularization that was presumably related to the disruption. Of 49 patients without acute disruption, late discontinuities were observed in 21 patients. There were no major adverse cardiac events associated with this finding except for 1 patient who had a non-ischemia-driven target lesion revascularization. CONCLUSIONS Acute scaffold disruption is a rare iatrogenic phenomenon that has been anecdotally associated with anginal symptoms, whereas late strut discontinuity is observed in approximately 40% of patients and could be viewed as a serendipitous OCT finding of a normal bioresorption process without clinical implications. (ABSORB Clinical Investigation, Cohort B [ABSORB B]; NCT00856856).