Bram Stieltjes

IVIM-diffusion-MRI for the differentiation of solid benign and malign hypervascular liver lesions—Evaluation with two different MR scanners

PURPOSE Hepatocellular carcinoma (HCC) and focal nodular hyperplasia (FNH) as the most common malign and benign liver tumors are both hypervascularized and potentially difficult to differentiate. DWI in liver MRI has been shown to be helpful in the classification of liver lesions, although with a substantial ADC-values-overlap. First results suggest that IVIM allows for improved characterization of liver lesions. In this study we evaluated IVIM-derived parameters in HCC and FNH with two different MR-scanners. MATERIAL AND METHODS 72 patients (29 FNH, 43HCC) were examined prospectively using two 1.5 T-MRI scanners (Aera/MagnetomAvanto, Siemens, Germany). Quantitative analysis of IVIM-derived parameters and ADC800-values was performed independently by two radiologists. The concordance between the reviewers was tested using a Pearson-/Spearman-correlation. The mean values for significant differences between FNHs and HCCs and between the two MR scanners were compared using a two-tailed t-test/Mann-Whitney-U test. An ROC analysis assessing the diagnostic performance of the parameters was performed. RESULTS The concordance between the two f-, D- and D*-measurements were r=0.81, 0.81 and 0.84, and r=0.58 for ADC-values. D-values and ADC800-values were significantly lower in HCC compared to FNH (p<0.001), there was no significant difference for f and D*. D had the largest AUC (0.76) for the differentiation between the two entities. Most parameters were not significantly different between the two MRIs. CONCLUSION IVIM-derived D and ADC are comparable for the differentiation between HCC and FNH. Since ADC-measurement means less effort than IVIM, ADC should be used for the differentiation between the two entities. Furthermore, quantitative results obtained from different scanners match closely.

The Bright, Artificial Intelligence-Augmented Future of Neuroimaging Reading

Radiologists are among the first physicians to be directly affected by advances in computer technology. Computers are already capable of analyzing medical imaging data, and with decades worth of digital information available for training, will an artificial intelligence (AI) one day signal the end of the human radiologist? With the ever increasing work load combined with the looming doctor shortage, radiologists will be pushed far beyond their current estimated 3 s allotted time-of-analysis per image; an AI with super-human capabilities might seem like a logical replacement. We feel, however, that AI will lead to an augmentation rather than a replacement of the radiologist. The AI will be relied upon to handle the tedious, time-consuming tasks of detecting and segmenting outliers while possibly generating new, unanticipated results that can then be used as sources of medical discovery. This will affect not only radiologists but all physicians and also researchers dealing with medical imaging. Therefore, we must embrace future technology and collaborate interdisciplinary to spearhead the next revolution in medicine.

Left atrial anatomy, atrial fibrillation burden, and P-wave duration—relationships and predictors for single-procedure success after pulmonary vein isolation

Aims Atrial fibrillation (AF) is associated with changes in left atrial (LA) volume, but the relationship between LA size, AF burden, and electrical conduction behaviour is still uncertain. The aim of this study was to quantify the association and impact of these parameters on the single-procedure outcome after circumferential antral ablation for pulmonary vein isolation. Methods and results Left atrial assessment was performed in 129 consecutive patients using pre-procedural imaging in three dimensions (sphericity, indexed volume), two dimensions (diameters), and from echocardiography in one dimension (long axis). Atrial fibrillation burden was classified based on the clinical assessment as paroxysmal and persistent and based on a validated scoring system including frequency, duration of AF episodes, and number of cardioversions into four grades (minimal, mild, moderate, and severe). P-wave duration and PR interval was measured on the 12-lead electrocardiogram at the end of the procedure. Atrial fibrillation burden score (AFB) was minimal (2%), mild (75%), moderate (9%), and severe (14%) and 65% had paroxysmal and 35% had persistent AF. The recurrence rate was significantly higher in patients with persistent AF, with higher AFB, with prolonged P-wave, and with an indexed LA volume  > 55 mL/m2. In multivariable analysis, AFB (hazard ratio: 2.018(1.383-2.945), P > 0.001) and a prolonged P-wave (hazard ratio: 2.612(1.248-5.466), P = 0.011) were identified as significant predictors for AF recurrence. Conclusions In our cohort of patients with symptomatic AF, the AFB and the P-wave duration but none of the anatomical parameter revealed to be independent predictors for AF/AT recurrence after circumferential antral pulmonary vein isolation.

Introduction to Diffusion Imaging

Diffusion is – for the purpose of this book – the random motion of water molecules in fluid water. When water is frozen, the particles stand still, but when water is liquid, the water molecules move around owing to their thermal energy. They do so rather quickly, at about 1,000 m/s. This above-racing-car speed, however, does not convert into a high-speed, long-distance distribution of the molecules, because they bounce at each other very frequently: The time between bounces is about 1/1,000,000,000,000 s only.

Comparison of image quality and radiation dose between split-filter dual-energy images and single-energy images in single-source abdominal CT

ObjectivesTo compare image quality and radiation dose of abdominal split-filter dual-energy CT (SF-DECT) combined with monoenergetic imaging to single-energy CT (SECT) with automatic tube voltage selection (ATVS).MethodsTwo-hundred single-source abdominal CT scans were performed as SECT with ATVS (n = 100) and SF-DECT (n = 100). SF-DECT scans were reconstructed and subdivided into composed images (SF-CI) and monoenergetic images at 55 keV (SF-MI). Objective and subjective image quality were compared among single-energy images (SEI), SF-CI and SF-MI. CNR and FOM were separately calculated for the liver (e.g. CNRliv) and the portal vein (CNRpv). Radiation dose was compared using size-specific dose estimate (SSDE). Results of the three groups were compared using non-parametric tests.ResultsImage noise of SF-CI was 18% lower compared to SEI and 48% lower compared to SF-MI (p < 0.001). Composed images yielded higher CNRliv over single-energy images (23.4 vs. 20.9; p < 0.001), whereas CNRpv was significantly lower (3.5 vs. 5.2; p < 0.001). Monoenergetic images overcame this inferiority in CNRpv and achieved similar results compared to single-energy images (5.1 vs. 5.2; p > 0.628). Subjective sharpness was equal between single-energy and monoenergetic images and diagnostic confidence was equal between single-energy and composed images. FOMliv was highest for SF-CI. FOMpv was equal for SEI and SF-MI (p = 0.78). SSDE was significant lower for SF-DECT compared to SECT (p < 0.022).ConclusionsThe combined use of split-filter dual-energy CT images provides comparable objective and subjective image quality at lower radiation dose compared to single-energy CT with ATVS.Key points• Split-filter dual-energy results in 18% lower noise compared to single-energy with ATVS.• Split-filter dual-energy results in 11% lower SSDE compared to single-energy with ATVS.• Spectral shaping of split-filter dual-energy leads to an increased dose-efficiency.

Überwachtes Lernen zur Prädiktion von Tumorwachstum

In der Bestrahlungsplanung bei Hirntumoren wird typischerweise ein Sicherheitsabstand von 2 − 2, 5 cm um das im T2-Flair MR-Bild hyperintense Gebiet eingeplant. Verlasliche Vorhersagen des Tumorwachstums konnen dazu beitragen, die Strahlendosis noch besser auf gefahrdete Regionen zu konzentrieren und gleichzeitig gesundes Gewebe zu schonen. Aktuelle Verfahren aus der Forschung nahern sich diesem Problem mit einer expliziten, generativen Modellierung des Wachstumsprozesses. Wir prasentieren ein alternatives, diskriminatives Verfahren. Mit Hilfe einer annotierten Datenbasis und uberwachtem Lernen wird ein Wachstumsmodell trainiert und im nachsten Schritt auf ungesehene Daten angewendet. In allen 6 Testpatienten lieferte der Ansatz genauere Vorhersagen (DICE 0, 80±0, 09) als die bisherige Herangehensweise (DICE 0, 56 ± 0, 07).

Automatische Tumorsegmentierung mit spärlich annotierter Lernbasis

Die Erstellung von Trainingsdaten fur lernbasierte Segmentierungsverfahren ist haufig sehr zeitaufwendig und fehleranfallig. Gleichzeitig muss die Lernbasis an die konkrete Bildgebung einer Klinik angepasst werden, was eine weite Verbreitung solcher automatischer Segmentierungsverfahren in der klinischen Routine verhindert. Wir schlagen daher ein Verfahren vor, welches durch die Verwendung eines Domain Adaption Ansatzes auf sparlichen, leicht anzufertigenden Segmentierungen trainiert werden kann. Wir validieren das vorgestellte System auf einem Kollektiv von 19 Patienten mit malignen Gliomen und zeigen, dass unser Ansatz die benotigte Annotierungszeit deutlich reduziert, wahrend die Klassifikationsergebnisse gegenuber klassisch trainierten Segmentierungsans atzen kaum beeintrachtigt werden. Der vorgestellte Ansatz erh oht die Attraktivitat automatischer Segmentierungsverfahren fur den klinischen Einsatz. Weiterhin lasst er die Erstellung umfangreicher Datenbanken mit grosen Fallzahlen fur unterschiedlichste Szenarien in greifbare Nahe rucken.

Navigierte ultraschallgeführte Leberpunktion mit integriertem EM Feldgenerator

Leberpunktionen sind ein elementares Werkzeug zur Diagnosesicherung von Raumforderungen. Zentrale Erfolgsfaktoren sind neben dem Treffen der Zielregion die Vermeidung von Risikostrukturen sowie eine geringe Eingriffsdauer. Es wurden bereits Navigationslosungen fur Ultraschall vorgeschlagen, welche aber aufgrund ihrer Komplexitat keine weite Verbreitung in der klinischen Praxis fanden. Wir stellen das erste Verfahren vor, welches eine Ultraschallsonde und einen neuen kompakten elektromagnetischen Feldgenerator zu einer einzigen mobilen Modalitat verbindet, mit welcher Patientenanatomie und Instrumente relativ zueinander erfasst werden konnen. In einer Phantomstudie zeigen wir, dass sich das neue Konzept fur eine akkurate Nadelinsertion ohne Verletzung von Risikostrukturen eignet.

Vorhersage des Krankheitsverlaufes von leichten kognitiven Beeinträchtigungen durch automatisierte MRT Morphometrie

Die leichte kognitive Beeintrachtigung (LKB) gilt als Anzeichen fur ein erhohtes Risiko der Entwicklung einer Alzheimerdemenz. Eine fundierte klinische Prognose fur den Krankheitsverlauf kann aber bis dato nicht gegeben werden. Das Ziel dieser Arbeit besteht darin, eine moglichst prazise Vorhersage mittels automatisierter Morphometrie des Hippokampus im MRT-Bild zu treffen. In einer Studie mit 18 Probanden mit LKB wurde eine Pradiktionsgenauigkeit fur die Entwicklung einer spateren Demenz von 83.3% erzielt. Eine manuelle Vergleichsmethode erreichte mit 55.6% Trefferquote keine signifikante Vorhersagegenauigkeit. Das automatische Verfahren erfullt viele wichtige Voraussetzungen fur den routinemasigen klinischen Einsatz mit dem Potential, die klinische Vorhersage des Krankheitsverlaufes bei der LKB zu verbessern.