Search

Biomedical image analysis algorithm validation depends on high-quality annotation of reference datasets, for which labelling instructions are key. Despite their importance, their optimization remains largely unexplored. Here we present a systematic study of labelling instructions and their impact on annotation quality in the field. Through comprehensive examination of professional practice and international competitions registered at the Medical Image Computing and Computer Assisted Intervention Society, the largest international society in the biomedical imaging field, we uncovered a discrepancy between annotators’ needs for labelling instructions and their current quality and availability. On the basis of an analysis of 14,040 images annotated by 156 annotators from four professional annotation companies and 708 Amazon Mechanical Turk crowdworkers using instructions with different information density levels, we further found that including exemplary images substantially boosts annotation performance compared with text-only descriptions, while solely extending text descriptions does not. Finally, professional annotators constantly outperform Amazon Mechanical Turk crowdworkers. Our study raises awareness for the need of quality standards in biomedical image analysis labelling instructions.

Allogeneic haematopoietic stem cell transplantation (alloSCT) often remains the only curative therapy for hematologic malignancies. Although the management of transplant-associated adverse events considerably improved over the last decades, nonrelapse mortality (NRM) remains a challenge, and endothelial dysfunction was identified as a major contributor to NRM.Statin-based endothelial prophylaxis (SEP) has been implemented in the standard of care in our transplant centre to reduce NRM caused by endothelial injury. Here, we retrospectively analysed the impact of SEP on clinical outcome in a cohort of 347 alloSCT patients.SEP (n = 209) was associated with significantly reduced NRM (hazard ratio 0.61, 95% CI 0.38-0.96) and better overall survival (OS) after acute graft-versus-host disease (HR 0.59, 95% CI 0.37-0.93). Subgroup analyses showed that the NRM benefit was mainly found in patients with an intermediate endothelial activation and stress index (EASIX), while relapse risk was not affected. On day 100 post-alloSCT, patients receiving SEP had significantly higher levels of the rate-limiting enzyme of tryptophan metabolism, indoleamine 2,3-dioxygenase (IDO), higher kynurenine to tryptophan ratios as a proxy of IDO activity and tended to have lower levels of the endothelial injury marker ST2 (p = .055). No significant differences in interferon-gamma or IL18 levels were observed. These biomarker signatures suggest that the beneficial effects of SEP might be mediated by both endothelial protection and immunomodulation.Together, these data suggest that SEP improves NRM and OS post-alloSCT in particular in patients with intermediate endothelial risk and provide first mechanistic clues about its potential mode of action.

Visual discrimination of tissue during surgery can be challenging since different tissues appear similar to the human eye. Hyperspectral imaging (HSI) removes this limitation by associating each pixel with high-dimensional spectral information. While previous work has shown its general potential to discriminate tissue, clinical translation has been limited due to the method’s current lack of robustness and generalizability. Specifically, it had been unknown whether variability in spectral reflectance is primarily explained by tissue type rather than the recorded individual or specific acquisition conditions. The contribution of this work is threefold: (1) Based on an annotated medical HSI data set (9,059 images from 46 pigs), we present a tissue atlas featuring spectral fingerprints of 20 different porcine organs and tissue types. (2) Using the principle of mixed model analysis, we show that the greatest source of variability related to HSI images is the organ under observation. (3) We show that HSI-based fully-automatic tissue differentiation of 20 organ classes with deep neural networks is possible with high accuracy (> 95 %). We conclude from our study that automatic tissue discrimination based on HSI data is feasible and could thus aid in intraoperative decision making and pave the way for context-aware computer-assisted surgery systems and autonomous robotics.Category for submissionExperimental studyLevel of EvidenceNot applicable. Translational animal science. Original article.

Laparoscopic surgery has evolved as a key technique for cancer diagnosis and therapy. While characterization of the tissue perfusion is crucial in various procedures, such as partial nephrectomy, doing so by means of visual inspection remains highly challenging. Spectral imaging takes advantage of the fact that different tissue components have unique optical properties to recover relevant information on tissue function such as ischemia. However, clinical success stories for advancing laparoscopic surgery with spectral imaging are lacking to date. To address this bottleneck, we developed the first laparoscopic real-time multispectral imaging (MSI) system featuring a compact and lightweight multispectral camera and the possibility to complement the conventional RGB (Red, Green, and Blue) surgical view of the patient with functional information at a video rate of 25 Hz. To account for the high inter-patient variability of human tissue, we phrase the problem of ischemia detection as an out-of-distribution (OoD) detection problem that does not rely on data from any other patient. Using an ensemble of invertible neural networks (INNs) as a core component, our algorithm computes the likelihood of ischemia based on a short (several seconds) video sequence acquired at the beginning of each surgery. A first-in-human trial performed on 10 patients undergoing partial nephrectomy demonstrates the feasibility of our approach for fully-automatic live ischemia monitoring during laparoscopic surgery. Compared to the clinical state-of-the-art approach based on indocyanine green (ICG) fluorescence, the proposed MSI-based method does not require the injection of a contrast agent and is repeatable if the wrong segment has been clamped. Spectral imaging combined with advanced deep learning-based analysis tools could thus evolve as an important tool for fast, efficient, reliable and safe functional imaging in minimally invasive surgery.

Lenalidomide and dexamethasone (RD) is a standard treatment in relapsed/refractory immunoglobulin light chain (AL) amyloidosis (RRAL). We retrospectively investigated toxicity, efficacy and prognostic markers in 260 patients with RRAL. Patients received a median of two prior treatment lines (68% had been bortezomib-refractory; 33% had received high-dose melphalan). The median treatment duration was four cycles. The 3-month haematological response rate was 31% [very good haematological response (VGHR) in 18%]. The median follow-up was 56·5 months and the median overall survival (OS) and haematological event-free survival (haemEFS) were 32 and 9 months. The 2-year dialysis rate was 15%. VGHR resulted in better OS (62 vs. 26 months, P < 0·001). Cardiac progression predicted worse survival (22 vs. 40 months, P = 0·027), although N-terminal prohormone of brain natriuretic peptide (NT-proBNP) increase was frequently observed. Multivariable analysis identified these prognostic factors: NT-proBNP for OS [hazard ratio (HR) 1·71; P < 0·001]; gain 1q21 for haemEFS (HR 1·68, P = 0·014), with a trend for OS (HR 1·47, P = 0·084); difference between involved and uninvolved free light chains (dFLC) and light chain isotype for OS (HR 2·22, P < 0·001; HR 1·62, P = 0·016) and haemEFS (HR 1·88, P < 0·001; HR 1·59, P = 0·008). Estimated glomerular filtration rate (HR 0·71, P = 0·004) and 24-h proteinuria (HR 1·10, P = 0·004) were prognostic for renal survival. In conclusion, clonal and organ biomarkers at baseline identify patients with favourable outcome, while VGHR and cardiac progression define prognosis during RD treatment.

Purpose: Endothelial complications are principal causes of non-relapse mortality (NRM) after allogeneic stem cell transplantation (alloSCT). Sepsis is a dysfunctional endothelial response to harmful microorganisms with increased risk of microvascular damage and organ failure. We hypothesized that the endothelial activation and stress index (EASIX) predicts risk of sepsis after alloSCT. Methods: In this retrospective evaluation, 1290 patients (random 1:1 allocation into a training cohort and a validation cohort while balancing for sepsis events) were assessed for presence of neutropenic fever, sepsis and infectious pathogens within 50 days after alloSCT. Sepsis and septic shock were defined according to the modified Sepsis-3 guidelines by the Infectious Diseases Working Party (AGIHO) and Intensive Care Working Party (iCHOP) of the German Society of Hematology and Medical Oncology (DGHO) for neutropenic cancer patients. EASIX and additional serum markers were assessed longitudinally before and after transplantation and correlated with outcome. Established clinical risk scores (EBMT, HCT-CI and VOD-CIBMTR) were raised and compared to EASIX. Results: Within the full cohort of 1290 patients transplanted since 01/2004, neutropenic fever episodes until day+50 after alloSCT were reported in 989 (76.7%), sepsis in 93 (7.2%), catecholamine use in 35 (2.7%), and mechanical ventilation in 31 (2.4%) patients. Patients who developed sepsis until day+50 differed from patients without early sepsis in more frequently having a higher disease score (58% vs 33%, p Prior to conditioning therapy, EASIX(-pre) was the only marker to predict the hazard of early sepsis irrespective of pathogen detection in time-dependent ROC (AUC 0.85 at day 50) and multivariable cause-specific Cox regression analyses (HR per two-fold increase 2.3 (1.9-2.8, p Pre-conditioning leukocyte counts, CRP or endothelial serum markers did not associate with early sepsis. After transplantation starting with day 0, sepsis associated with EASIX at any time point until day+28, and in addition with suppressor of tumorigenicity (ST)2 and interleukin-18. Conclusions: EASIX is a powerful predictor of early sepsis and 6-months NRM, irrespective of detected pathogens. Patients with EASIX-pre>2.32 represent a high-risk cohort requiring intensive prophylaxis and endothelial protective treatment strategies. Figure 1 Figure 1. Disclosures Schönland: Pfizer: Honoraria; Janssen: Honoraria, Other: Travel grants, Research Funding; Prothena: Honoraria, Other: Travel grants; Sanofi: Research Funding; Takeda: Honoraria, Other: Travel grants. Hegenbart: Pfizer: Consultancy, Honoraria; Janssen: Consultancy, Research Funding; Alnylam: Honoraria; Akcea: Honoraria; Prothena: Research Funding. Schmitt: Apogenix: Research Funding; Hexal: Other: Travel grants, Research Funding; Novartis: Other: Travel grants, Research Funding; Kite Gilead: Other: Travel grants; Bluebird Bio: Other: Travel grants; MSD: Membership on an entity's Board of Directors or advisory committees; TolerogenixX: Current holder of individual stocks in a privately-held company. Dreger: Novartis: Consultancy, Speakers Bureau; Bluebird Bio: Consultancy; Janssen: Consultancy; Gilead Sciences: Consultancy, Speakers Bureau; BMS: Consultancy; AbbVie: Consultancy, Speakers Bureau; AstraZeneca: Consultancy, Speakers Bureau; Riemser: Consultancy, Research Funding, Speakers Bureau; Roche: Consultancy, Speakers Bureau.

The additive genomic variance in linear models with random marker effects can be defined as a random variable that is in accordance with classical quantitative genetics theory. Common approaches to estimate the genomic variance in random-effects linear models based on genomic marker data can be regarded as estimating the unconditional (or prior) expectation of this random additive genomic variance, and result in a negligence of the contribution of linkage disequilibrium (LD). We introduce a novel best prediction (BP) approach for the additive genomic variance in both the current and the base population in the framework of genomic prediction using the genomic best linear unbiased prediction (gBLUP) method. The resulting best predictor is the conditional (or posterior) expectation of the additive genomic variance when using the additional information given by the phenotypic data, and is structurally in accordance with the genomic equivalent of the classical additive genetic variance in random-effects models. In particular, the best predictor includes the contribution of (marker) LD to the additive genomic variance and possibly fully eliminates the missing contribution of LD that is caused by the assumptions of statistical frameworks such as the random-effects model. We derive an empirical best predictor (eBP) and compare its performance with common approaches to estimate the additive genomic variance in random-effects models on commonly used genomic datasets.