No description available.

Capsule Endoscopy (CE) is a unique technique for facilitating non-invasive and practical visualization of the entire small intestine. It has attracted a critical mass of studies for improvements. Among numerous studies being performed in capsule endoscopy, tremendous efforts are being made in the development of software algorithms to identify clinically important frames in CE videos. This thesis presents a computer-assisted method which performs automated detection of CE video-frames that contain bleeding. Specifically, a methodology is proposed to classify the frames of CE videos into bleeding and non-bleeding frames. It is a Support Vector Machine (SVM) based supervised method which classifies the frames on the basis of color features derived from image-regions. Image-regions are characterized on the basis of statistical features. With 15 available candidate features, an exhaustive feature-selection is followed to obtain the best feature subset. The best feature-subset is the combination of features that has the highest bleeding discrimination ability as determined by the three performance-metrics: accuracy, sensitivity and specificity. Also, a ground truth label annotation method is proposed in order to partially automate delineation of bleeding regions for training of the classifier. The method produced promising results with sensitivity and specificity values up to 94%. All the experiments were performed separately for RGB and HSV color spaces. Experimental results show the combination of the mean planes in red and green planes to be the best feature-subset in RGB (Red-Green-Blue) color space and the combination of the mean values of all three planes of the color space to be the best feature-subset in HSV (Hue-Saturation-Value).

Capsule Endoscopy (CE) is a unique technique for facilitating non-invasive and practical visualization of the entire small intestine. It has attracted a critical mass of studies for improvements. Among numerous studies being performed in capsule endoscopy, tremendous efforts are being made in the development of software algorithms to identify clinically important frames in CE videos. This thesis presents a computer-assisted method which performs automated detection of CE video-frames that contain bleeding. Specifically, a methodology is proposed to classify the frames of CE videos into bleeding and non-bleeding frames. It is a Support Vector Machine (SVM) based supervised method which classifies the frames on the basis of color features derived from image-regions. Image-regions are characterized on the basis of statistical features. With 15 available candidate features, an exhaustive feature-selection is followed to obtain the best feature subset. The best feature-subset is the combination of features that has the highest bleeding discrimination ability as determined by the three performance-metrics: accuracy, sensitivity and specificity. Also, a ground truth label annotation method is proposed in order to partially automate delineation of bleeding regions for training of the classifier. The method produced promising results with sensitivity and specificity values up to 94%. All the experiments were performed separately for RGB and HSV color spaces. Experimental results show the combination of the mean planes in red and green planes to be the best feature-subset in RGB (Red-Green-Blue) color space and the combination of the mean values of all three planes of the color space to be the best feature-subset in HSV (Hue-Saturation-Value).

Lange Zeit galt der Dünndarm als „Blackbox“ des Gastrointestinaltraktes. Seit Einführung der Videokapselendoskopie im Jahr 2001 eröffnete sich eine Methode, den Dünndarm zu visualisieren. An einem Kollektiv von 203 Patienten habe ich Indikationen, Ergebnisse und klinischen Nutzen von Dünndarmkapselendoskopien in einem Zeitraum von 4 Jahren untersucht. Der Dünndarm ist in der Gastroduodeno- und Koloskopie nicht komplett zugänglich, so dass bei entsprechender Indikation die nicht-invasive Videokapselendoskopie vorgenommen werden kann. Sie ist in der Lage 2-4 Bilder pro Sekunde in einem Zeitraum von 8-9 Stunden aufzunehmen, die als Film von ca. 50.000 Bildern zusammengestellt und interpretiert werden kann. Die Daten zur diagnostischen Ausbeute dieser Untersuchung variieren und sind abhängig von der entsprechenden Indikation. Zur Überprüfung des klinischen Nutzens habe ich daher in meiner Arbeit speziell die Passagezeiten und die erhobenen Befunde, wie Erosionen, Ulzerationen, Angiodysplasien, Petechien, Venektasien, Lymphangiektasien, Erytheme, Ödeme, Zottenreliefveränderungen, extrinsische Engen und Erhabenheiten im Hinblick für ihre diagnostische Bedeutung ausgewertet. Berücksichtigt wurden die Auswertbarkeit, Komplikationsrate sowie Vor- und Nachuntersuchungen. Das Aufklärungsgespräch erfolgte mindestens einen Tag vor der Videokapselendoskopie. Die Abführmaßnahmen entsprachen einer Koloskopievorbereitung. Das Studienkollektiv (203 Patienten) bestand aus 58% männlichen und 42% weiblichen Patienten. Der Altersdurchschnitt betrug 58 Jahre, die Altersspanne reichte von 8-90 Jahren. Über 93% nahmen die Videokapsel selbstständig ein, eine Applikation erfolgte bei 7% der Patienten in den Bulbus duodeni. Folgende Indikationen führten bei unserer Patientenklientel zu der Videokapselendoskopie: unklare gastrointestinale Blutung (45,3%), unklare abdominelle Schmerzen (24,1%), unklare Anämie (11,3%), Verdacht auf/ oder Komplikation bei Morbus Crohn (6,5%), unklare Diarrhoe (6,4%), Polyp- und Tumorsuche (5,4%), rezidivierendes unklares Erbrechen und Eiweißverlustsyndrom (jeweils 0,5%). Eine komplette Dünndarmpassage konnte innerhalb der Aufzeichnungszeit von 8-9 Stunden bei 84% der Patienten erreicht werden. Der Mittelwert der Magenpassagezeit lag bei 21 Minuten und der Dünndarmpassagezeit bei 6 Stunden. Die Komplikation Kapselretention trat bei 2% auf. Pathologische Befunde im Dünndarm wurden bei 85% detektiert. Die höchste diagnostische Ausbeute ergab sich bei der Abklärung der unklaren gastrointestinalen Blutung (80%) und bei der unklaren Anämie (78%), als häufigste Ursache wurden Schleimhautläsionen (43%) gefunden. Unklare abdominelle Schmerzen wiesen eine niedrigere diagnostische Ausbeute (41%) auf. Therapeutische Maßnahmen resultierten bei 73% der untersuchten Patienten aus den Kapselergebnissen. Eine medikamentöse Therapie wurde bei 66% eingeleitet oder verändert, Endoskopien wurden bei 4% und eine operative Therapie bei 4,4% durchgeführt. Damit ist die Dünndarmkapselendoskopie bei klarer Fragestellung und guter Darmvorbereitung eine sichere und sinnvolle Untersuchungsmethode, insbesondere zur Klärung unklarer gastrointestinaler Blutungen. Spezifische Dünndarmerkrankungen, wie der M. Crohn oder Tumore können relativ sicher ausgeschlossen werden.

BACKGROUND: Capsule endoscopy (CE) allows for a non-invasive small bowel evaluation for a wide range of gastrointestinal (GI) symptoms and diseases. Capsule technology has been rapidly advancing over recent years, often improving image frequency and quality. The Pillcam SB3 (SB3) capsule is one such technology that offers an adaptive frame rate advantage over the previous versions of the capsule the Pillcam SB2 (SB2). Some have proposed that this improvement in capsule technology may lead to increased diagnostic yields; however, real world clinical data is currently lacking. AIM: To evaluate the clinically relevant findings of SB3 and SB2 capsules in a population of United States veterans. METHODS: A retrospective analysis of 260 consecutive CE studies was performed including 130 SB3 and 130 SB2 capsule studies. Recorded variables included: age, gender, type of capsule, body mass index, exam completion, inpatient status, opioid use, diabetes, quality of preparation, gastric transit time, small bowel transit time, indication, finding, and if the exam resulted in a change in clinical management. The primary outcome measured was the detection of clinically relevant findings between SB3 and SB2 capsules. RESULTS: Mean age of the study population was 67.1 ± 10.4 years and 94.2% of patients were male. Of these 28.1% were on opioid users. The most common indications for capsule procedure were occult GI bleeding (74.6%) and overt GI bleeding (14.6%). Rates of incomplete exam were similar between SB3 and SB2 groups (16.9% 9.2%, = 0.066). The overall rate of clinically relevant finding was 48.9% in our study. No significant difference was observed in SB3 SB2 capsules for clinically relevant findings (46.2% 51.5%, = 0.385) or change in clinical management (40.8% 50.0%, = 0.135). CONCLUSION: Our study found no significant difference in clinically relevant findings between SB3 and SB2 capsules.

Yes / Video capsule endoscopy (VCE) is a revolutionary technology for the early diagnosis of gastric disorders. However, owing to the high redundancy and subtle manifestation of anomalies among thousands of frames, the manual construal of VCE videos requires considerable patience, focus, and time. The automatic analysis of these videos using computational methods is a challenge as the capsule is untamed in motion and captures frames inaptly. Several machine learning (ML) methods, including recent deep convolutional neural networks approaches, have been adopted after evaluating their potential of improving the VCE analysis. However, the clinical impact of these methods is yet to be investigated. This survey aimed to highlight the gaps between existing ML-based research methodologies and clinically significant rules recently established by gastroenterologists based on VCE. A framework for interpreting raw frames into contextually relevant frame-level findings and subsequently merging these findings with meta-data to obtain a disease-level diagnosis was formulated. Frame-level findings can be more intelligible for discriminative learning when organized in a taxonomical hierarchy. The proposed taxonomical hierarchy, which is formulated based on pathological and visual similarities, may yield better classification metrics by setting inference classes at a higher level than training classes. Mapping from the frame level to the disease level was structured in the form of a graph based on clinical relevance inspired by the recent international consensus developed by domain experts. Furthermore, existing methods for VCE summarization, classification, segmentation, detection, and localization were critically evaluated and compared based on aspects deemed significant by clinicians. Numerous studies pertain to single anomaly detection instead of a pragmatic approach in a clinical setting. The challenges and opportunities associated with VCE analysis were delineated. A focus on maximizing the discriminative power of features corresponding to various subtle lesions and anomalies may help cope with the diverse and mimicking nature of different VCE frames. Large multicenter datasets must be created to cope with data sparsity, bias, and class imbalance. Explainability, reliability, traceability, and transparency are important for an ML-based diagnostics system in a VCE. Existing ethical and legal bindings narrow the scope of possibilities where ML can potentially be leveraged in healthcare. Despite these limitations, ML based video capsule endoscopy will revolutionize clinical practice, aiding clinicians in rapid and accurate diagnosis.

The early discovery of gastrointestinal (GI) disorders can significantly decrease the fatality rate of severe afflictions. Video capsule endoscopy (VCE) is a technique that produces an eight hour long recording of the GI tract that needs to be manually reviewed. This has led to the demand for AI-based solutions, but unfortunately, the lack of labeled data has been a major obstacle. In 2020 the Kvasir-Capsule dataset was produced which is the largest labeled dataset of GI abnormalities to date, but challenges still exist.The dataset suffers from unbalanced and very similar data created from labeled video frames. To avoid specialization to the specific data the creators of the set constructed an official split which is encouraged to use for testing. This study evaluates the use of transfer learning, Data augmentation and binary classification to detect GI abnormalities. The performance of machine learning (ML) classification is explored, with and without official split-based testing. For the performance evaluation, a specific focus will be on achieving a low rate of false negatives. The proposition behind this is that the most important aspect of an automated detection system for GI abnormalities is a low miss rate of possible lethal abnormalities. The results from the controlled experiments conducted in this study clearly show the importance of using official split-based testing. The difference in performance between a model trained and tested on the same set and a model that uses official split-based testing is significant. This enforces that without the use of official split-based testing the model will not produce reliable and generalizable results. When using official split-based testing the performance is improved compared to the initial baseline that is presented with the Kvasir-Capsule set. Some experiments in the study produced results with as low as a 1.56% rate of false negatives but with the cost of lowered performance for the normal class.

Dieser Beitrag ist mit Zustimmung des Rechteinhabers aufgrund einer (DFG-geförderten) Allianz- bzw. Nationallizenz frei zugänglich.

[ES] La cápsula inalámbrica de endoscopia (WCE) es una tecnología notable y atractiva adoptada en el sector biomédico hace varios años. WCE proporciona una tecnología de imagen inalámbrica no invasiva que permite a los especialistas reconocer y diagnosticar enfermedades que afectan todo el tracto gastrointestinal. Aunque los médicos pueden recibir imágenes claras de anomalías en el tracto gastrointestinal, no tienen información sobre sus exacta ubicación. La localización precisa de los trastornos detectados es crucial para el posterior procedimiento de extracción mediante cirugía. Actualmente, la banda de frecuencia asignada para aplicaciones de cápsula endoscópica es la banda MICS (402-405 MHz) que ofrece una velocidad de datos de hasta 500 kbps, insuciente para transmitir imágenes de alta calidad. Recientemente, la tecnología de banda ultra ancha (UWB) ha estado atrayendo atención como posible candidato para la próxima generación de cápsula endoscópica. Las ventajas de UWB incluyen arquitecturas de transceptor simples que permiten bajo consumo de potencia, baja interferencia a otros sistemas y amplio ancho de banda que resulta en comunicaciones a una velocidad de datos más alta. En esta disertación, el rendimiento de las técnicas de localización de WCE basadas en radiofrecuencia (RF) se investiga a través de simulaciones software, medidas experimentales de laboratorio que involucran fantomas homogéneos y heterogéneos y a través de experimentos in vivo que constituyen el escenario de prueba más realista. La tecnología UWB (3.1-10.6 GHz) se considera como interfaz de comunicación para aplicaciones de cápsula endoscópica. En tal escenario, el transmisor inalámbrico está ubicado en el tracto gastrointestinal, mientras que uno o más receptores inalámbricos están ubicados sobre la supercie del cuerpo. El enfoque basado en la potencia recibida (RSS) se investiga principalmente debido a su simplicidad de implementación y menos sensibilidad a las limitaciones de ancho de banda. Se analiza el impacto de la posición y del número de receptores seleccionados en la precisión de la localización. Finalmente, se desarrolla una interfaz gráfica de usuario (GUI) para visualizar los resultados de la localización en tres dimensiones (3D) obtenidos mediante las medidas in vivo. / [CAT] La càpsula sense fil d'endoscòpia (WCE) és una tecnologia notable i atractiva adoptada en el sector biomèdic fa diversos anys. La WCE proporciona una tecnologia d'imatge sense fil no invasiva que permet als especialistes reconéixer i diagnosticar malalties que afecten tot el tracte gastrointestinal. Encara que els metges poden rebre imatges clares d'anomalies en el tracte gastrointestinal, no tenen informació sobre les seues exacta ubicació. La localització precisa dels trastorns detectats és crucial per al posterior procediment d'extracció mitjançant cirurgia. Actualment, la banda de freqüència assignada per a aplicacions de càpsula endoscòpica és la banda MICS (402-405 MHz) que ofereix una velocitat de dades de fins a 500 kbps, insucient per a transmetre imatges d'alta qualitat. Recentment, la tecnologia de banda ultra ampla (UWB) ha estat atraient atenció com a possible candidata per a la pròxima generació de càpsula endoscòpica. Els avantatges d' UWB inclouen arquitectures de transceptor simples que permeten un baix consum de potència, baixa interferència amb altres sistemes i una gran amplada de banda que resulta en comunicacions a una velocitat de dades més alta. En aquesta dissertació, el rendiment de les tècniques de localització de WCE basades en radiofrequència (RF) s'investiga a través de simulacions amb programari, mesures experimentals de laboratori que involucren fantomes homogenis i heterogenis i a través d'experiments in vivo que constitueixen l'escenari de prova més realista. La tecnologia UWB (3.1-10.6 GHz) es considera com a interfície de comunicació per a aplicacions de càpsula endoscòpica. En tal escenari, el transmissor sense fil està situat en el tracte gastrointestinal, mentre que un o més receptors sense fils estan situats sobre la superfície del cos. L'enfocament basat en la potència rebuda (RSS) s'investiga principalment a causa de la seua simplicitat d'implementació i menys sensibilitat a les limitacions d'amplada de banda. S'analitza l'impacte de la posició i del numere de receptors seleccionats en la precisió de la localització. Finalment, es desenvolupa una interfície gràca d'usuari (GUI) per a visualitzar els resultats de la localització en tres dimensions (3D) obtinguts mitjançant les mesures in vivo. / [EN] Wireless Capsule Endoscopy (WCE) is a remarkable and attractive technology adopted in the biomedical sector several years ago. It provides a non-invasive wireless imaging technology for the entire gastrointestinal (GI) tract. WCE allows specialists to recognize and diagnose diseases affecting the whole GI tract. Although physicians can receive clear pictures of abnormalities in the GI tract, they have no information about their exact location. Precise localization of the detected disorders is crucial for the subsequent removal procedure by surgery. Currently, the frequency band allocated for capsule endoscopy applications is the MICS band (402-405 MHz). This band offers data rate up to 500 kbps, which is insufficient to transmit high quality images. Recently, Ultrawideband (UWB) technology has been attracting attention as potential candidate for next-generation WCE systems. The advantages of UWB include simple transceiver architectures enabling low power consumption, low interference to other systems and wide bandwidth resulting in communications at higher data rate. In this dissertation, performance of WCE localization techniques based on Radio Frequency (RF) information are investigated through software simulations, experimental laboratory measurements involving homogeneous and heterogeneous phantom models and in vivo experiments which constitute the most realistic testing scenario. Ultra-Wideband technology (3.1-10.6 GHz) is considered as communication interface in Wireless Capsule Endoscopy. In such scenario, the wireless transmitter is located in the gastrointestinal track while one or more wireless receivers are located over the surface of the body. Received Signal Strength (RSS)-based approach is mainly explored due to its implementation simplicity and less sensitivity to bandwidth limitations. Impact of the position and the number of selected receivers on the localization accuracy is analyzed. Finally, a graphical user interface (GUI) is developed to visualize the three-dimensional (3D) localization results obtained through in vivo measurements. / Barbi, M. (2019). Location and Tracking for Ultra-WideBand In-Body Communications in Medical Applications [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/132874 / TESIS

Dieser Beitrag ist mit Zustimmung des Rechteinhabers aufgrund einer (DFG-geförderten) Allianz- bzw. Nationallizenz frei zugänglich.