| 1. |
- Berlin, Jonas, et al.
(författare)
-
Trajectory Generation for Mobile Robots in a Dynamic Environment using Nonlinear Model Predictive Control
- 2021
-
Ingår i: IEEE International Conference on Automation Science and Engineering. - 2161-8070 .- 2161-8089. ; 2021-August, s. 942-947
-
Konferensbidrag (refereegranskat)abstract
- This paper presents an approach to collision-free, long-range trajectory generation for a mobile robot in an industrial environment with static and dynamic obstacles. For the long-range planning a visibility graph together with A is used to find a collision-free path with respect to the static obstacles. This path is used as a reference path to the trajectory planning algorithm that in addition handles dynamic obstacles while complying with the robot dynamics and constraints. A Nonlinear Model Predictive Control (NMPC) solver generates a collision-free trajectory by staying close to the initial path but at the same time obeying all constraints. The NMPC problem is solved efficiently by leveraging the new numerical optimization method Proximal Averaged Newton for Optimal Control (PANOC). The algorithm was evaluated by simulation in various environments and successfully generated feasible trajectories spanning hundreds of meters in a tractable time frame.
|
|
| 2. |
- Condoluci, Adalgisa, et al.
(författare)
-
International prognostic score for asymptomatic early-stage chronic lymphocytic leukemia
- 2020
-
Ingår i: Blood. - : American Society of Hematology. - 0006-4971 .- 1528-0020. ; 135:21, s. 1859-1869
-
Tidskriftsartikel (refereegranskat)abstract
- Most patients with chronic lymphocytic leukemia (CLL) are diagnosed with early-stage disease and managed with active surveillance. The individual course of patients with early-stage CLL is heterogeneous, and their probability of needing treatment is hardly anticipated at diagnosis. We aimed at developing an international prognostic score to predict time to first treatment (TTFT) in patients with CLL with early, asymptomatic disease (International Prognostic Score for Early-stage CLL [IPS-E]). Individual patient data from 11 international cohorts of patients with early-stage CLL (n = 4933) were analyzed to build and validate the prognostic score. Three covariates were consistently and independently correlated with TTFT: unmutated immunoglobulin heavy variable gene (IGHV), absolute lymphocyte count higher than 15 x 10(9)/L, and presence of palpable lymph nodes. The IPS-E was the sum of the covariates (1 point each), and separated low-risk (score 0), intermediate-risk (score 1), and high-risk (score 2-3) patients showing a distinct TTFT. The score accuracy was validated in 9 cohorts staged by the Binet system and 1 cohort staged by the Rai system. The C-index was 0.74 in the training series and 0.70 in the aggregate of validation series. By meta-analysis of the training and validation cohorts, the 5-year cumulative risk for treatment start was 8.4%, 28.4%, and 61.2% among low-risk, intermediate-risk, and high-risk patients, respectively. The IPS-E is a simple and robust prognostic model that predicts the likelihood of treatment requirement in patients with early-stage CLL. The IPS-E can be useful in clinical management and in the design of early intervention clinical trials.
|
|
| 3. |
- Dreyling, Martin, et al.
(författare)
-
Ibrutinib combined with immunochemotherapy with or without autologous stem-cell transplantation versus immunochemotherapy and autologous stem-cell transplantation in previously untreated patients with mantle cell lymphoma (TRIANGLE) : a three-arm, randomised, open-label, phase 3 superiority trial of the European Mantle Cell Lymphoma Network
- 2024
-
Ingår i: The Lancet. - 0140-6736. ; 403:10441, s. 2293-2306
-
Tidskriftsartikel (refereegranskat)abstract
- Background: Adding ibrutinib to standard immunochemotherapy might improve outcomes and challenge autologous stem-cell transplantation (ASCT) in younger (aged 65 years or younger) mantle cell lymphoma patients. This trial aimed to investigate whether the addition of ibrutinib results in a superior clinical outcome compared with the pre-trial immunochemotherapy standard with ASCT or an ibrutinib-containing treatment without ASCT. We also investigated whether standard treatment with ASCT is superior to a treatment adding ibrutinib but without ASCT. Methods: The open-label, randomised, three-arm, parallel-group, superiority TRIANGLE trial was performed in 165 secondary or tertiary clinical centres in 13 European countries and Israel. Patients with previously untreated, stage II–IV mantle cell lymphoma, aged 18–65 years and suitable for ASCT were randomly assigned 1:1:1 to control group A or experimental groups A+I or I, stratified by study group and mantle cell lymphoma international prognostic index risk groups. Treatment in group A consisted of six alternating cycles of R-CHOP (intravenous rituximab 375 mg/m2 on day 0 or 1, intravenous cyclophosphamide 750 mg/m2 on day 1, intravenous doxorubicin 50 mg/m2 on day 1, intravenous vincristine 1·4 mg/m2 on day 1, and oral prednisone 100 mg on days 1–5) and R-DHAP (or R-DHAOx, intravenous rituximab 375 mg/m2 on day 0 or 1, intravenous or oral dexamethasone 40 mg on days 1–4, intravenous cytarabine 2 × 2 g/m2 for 3 h every 12 h on day 2, and intravenous cisplatin 100 mg/m2 over 24 h on day 1 or alternatively intravenous oxaliplatin 130 mg/m2 on day 1) followed by ASCT. In group A+I, ibrutinib (560 mg orally each day) was added on days 1–19 of R-CHOP cycles and as fixed-duration maintenance (560 mg orally each day for 2 years) after ASCT. In group I, ibrutinib was given the same way as in group A+I, but ASCT was omitted. Three pairwise one-sided log-rank tests for the primary outcome of failure-free survival were statistically monitored. The primary analysis was done by intention-to-treat. Adverse events were evaluated by treatment period among patients who started the respective treatment. This ongoing trial is registered with ClinicalTrials.gov, NCT02858258. Findings: Between July 29, 2016 and Dec 28, 2020, 870 patients (662 men, 208 women) were randomly assigned to group A (n=288), group A+I (n=292), and group I (n=290). After 31 months median follow-up, group A+I was superior to group A with 3-year failure-free survival of 88% (95% CI 84–92) versus 72% (67–79; hazard ratio 0·52 [one-sided 98·3% CI 0–0·86]; one-sided p=0·0008). Superiority of group A over group I was not shown with 3-year failure-free survival 72% (67–79) versus 86% (82–91; hazard ratio 1·77 [one-sided 98·3% CI 0–3·76]; one-sided p=0·9979). The comparison of group A+I versus group I is ongoing. There were no relevant differences in grade 3–5 adverse events during induction or ASCT between patients treated with R-CHOP/R-DHAP or ibrutinib combined with R-CHOP/R-DHAP. During maintenance or follow-up, substantially more grade 3–5 haematological adverse events and infections were reported after ASCT plus ibrutinib (group A+I; haematological: 114 [50%] of 231 patients; infections: 58 [25%] of 231; fatal infections: two [1%] of 231) compared with ibrutinib only (group I; haematological: 74 [28%] of 269; infections: 52 [19%] of 269; fatal infections: two [1%] of 269) or after ASCT (group A; haematological: 51 [21%] of 238; infections: 32 [13%] of 238; fatal infections: three [1%] of 238). Interpretation: Adding ibrutinib to first-line treatment resulted in superior efficacy in younger mantle cell lymphoma patients with increased toxicity when given after ASCT. Adding ibrutinib during induction and as maintenance should be part of first-line treatment of younger mantle cell lymphoma patients. Whether ASCT adds to an ibrutinib-containing regimen is not yet determined. Funding: Janssen and Leukemia & Lymphoma Society.
|
|
| 4. |
- Hess, Georg, 1996, et al.
(författare)
-
LidarCLIP or: How I Learned to Talk to Point Clouds
- 2024
-
Ingår i: Proceedings of the 2024 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV). - 2642-9381. - 9798350318920 ; , s. 7423-7432
-
Konferensbidrag (refereegranskat)abstract
- Research connecting text and images has recently seen several breakthroughs, with models like CLIP, DALL•E 2, and Stable Diffusion. However, the connection between text and other visual modalities, such as lidar data, has received less attention, prohibited by the lack of text-lidar datasets. In this work, we propose LidarCLIP, a mapping from automotive point clouds to a pre-existing CLIP embedding space. Using image-lidar pairs, we supervise a point cloud encoder with the image CLIP embeddings, effectively relating text and lidar data with the image domain as an intermediary. We show the effectiveness of Lidar-CLIP by demonstrating that lidar-based retrieval is generally on par with image-based retrieval, but with complementary strengths and weaknesses. By combining image and lidar features, we improve upon both single-modality methods and enable a targeted search for challenging detection scenarios under adverse sensor conditions. We also explore zero-shot classification and show that LidarCLIP outperforms existing attempts to use CLIP for point clouds by a large margin. Finally, we leverage our compatibility with CLIP to explore a range of applications, such as point cloud captioning and lidar-to-image generation, without any additional training. Code and pre-trained models at github.com/atonderski/lidarclip.
|
|
| 5. |
- Hess, Georg, 1996, et al.
(författare)
-
Masked Autoencoder for Self-Supervised Pre-Training on Lidar Point Clouds
- 2023
-
Ingår i: Proceedings - 2023 IEEE/CVF Winter Conference on Applications of Computer Vision Workshops, WACVW 2023. ; , s. 350-359
-
Konferensbidrag (refereegranskat)abstract
- Masked autoencoding has become a successful pretraining paradigm for Transformer models for text, images, and, recently, point clouds. Raw automotive datasets are suitable candidates for self-supervised pre-training as they generally are cheap to collect compared to annotations for tasks like 3D object detection (OD). However, the development of masked autoencoders for point clouds has focused solely on synthetic and indoor data. Consequently, existing methods have tailored their representations and models toward small and dense point clouds with homogeneous point densities. In this work, we study masked autoencoding for point clouds in an automotive setting, which are sparse and for which the point density can vary drastically among objects in the same scene. To this end, we propose Voxel-MAE, a simple masked autoencoding pre-training scheme designed for voxel representations. We pre-train the backbone of a Transformer-based 3D object detector to reconstruct masked voxels and to distinguish between empty and non-empty voxels. Our method improves the 3D OD performance by 1.75 mAP points and 1.05 NDS on the challenging nuScenes dataset. Further, we show that by pre-training with Voxel-MAE, we require only 40 of the annotated data to outperform a randomly initialized equivalent. Code is available at https://github.com/georghess/voxel-mae.
|
|
| 6. |
- Hess, Georg, 1996, et al.
(författare)
-
Object Detection as Probabilistic Set Prediction
- 2022
-
Ingår i: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). - Cham : Springer Nature Switzerland. - 1611-3349 .- 0302-9743. - 9783031200793 ; 13670:XVIII, s. 550-566
-
Konferensbidrag (refereegranskat)abstract
- Accurate uncertainty estimates are essential for deploying deep object detectors in safety-critical systems. The development and evaluation of probabilistic object detectors have been hindered by shortcomings in existing performance measures, which tend to involve arbitrary thresholds or limit the detector’s choice of distributions. In this work, we propose to view object detection as a set prediction task where detectors predict the distribution over the set of objects. Using the negative log-likelihood for random finite sets, we present a proper scoring rule for evaluating and training probabilistic object detectors. The proposed method can be applied to existing probabilistic detectors, is free from thresholds, and enables fair comparison between architectures. Three different types of detectors are evaluated on the COCO dataset. Our results indicate that the training of existing detectors is optimized toward non-probabilistic metrics. We hope to encourage the development of new object detectors that can accurately estimate their own uncertainty. Code at https://github.com/georghess/pmb-nll.
|
|
| 7. |
- Hess, Georg, et al.
(författare)
-
Real-world experience among patients with relapsed/refractory mantle cell lymphoma after Bruton tyrosine kinase inhibitor failure in Europe : The SCHOLAR-2 retrospective chart review study
- 2023
-
Ingår i: British Journal of Haematology. - : Wiley. - 0007-1048 .- 1365-2141. ; 202:4, s. 749-759
-
Tidskriftsartikel (refereegranskat)abstract
- Mantle cell lymphoma (MCL) after relapse is associated with poor prognosis. No standard of care exists and available evidence for treatments is limited, particularly in patients who fail Bruton tyrosine kinase inhibitor (BTKi) therapy. This multicentre retrospective chart review study, SCHOLAR-2, addresses this knowledge gap and reports on data collected from 240 patients with relapsed/refractory MCL in Europe who were treated with BTKi-based therapy between July 2012 and July 2018, and had experienced disease progression while on BTKi therapy or discontinued BTKi therapy due to intolerance. The median overall survival (OS) from initiation of first BTKi therapy was 14.6 months (95% confidence interval [CI] 11.6–20.0) in the overall cohort, 5.5 months (95% CI 3.9–8.2) in 91 patients without post-BTKi therapy, and 23.8 months (95% CI 18.9–30.1) in 149 patients who received post-BTKi therapy (excluding chimeric antigen receptor T-cell treatment). In the latter group, patients received a median of one (range, one to seven) line of post-BTKi therapy, with lenalidomide-containing regimens and bendamustine plus rituximab being the most frequently administered; the median OS from initiation of first post-BTKi therapy was 9.7 months (95% CI 6.3–12.7). These results provide a benchmark for survival in patients with R/R MCL receiving salvage therapy after BTKi failure.
|
|
| 8. |
- Pinto, Juliano, 1990, et al.
(författare)
-
Deep Learning for Model-Based Multi-Object Tracking
- 2023
-
Ingår i: IEEE Transactions on Aerospace and Electronic Systems. - : IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC. - 1557-9603 .- 0018-9251. ; 59:6, s. 7363-7379
-
Tidskriftsartikel (refereegranskat)abstract
- Multi-object tracking (MOT) is the problem of tracking the state of an unknown and time-varying number of objects using noisy measurements, with important applications such as autonomous driving, tracking animal behavior, defense systems, and others. The MOT task can be divided into two settings, model-based or model-free, depending on whether accurate and tractable models of the environment are available. Model-based MOT has Bayes-optimal closed-form solutions which can achieve state-of-the-art (SOTA) performance. However, these methods require approximations in challenging scenarios to remain tractable, which impairs their performance. Deep learning (DL) methods offer a promising alternative, but existing DL models are almost exclusively designed for a model-free setting and are not easily translated to the model-based setting. This paper proposes a DL-based tracker specifically tailored to the model-based MOT setting and provides a thorough comparison to SOTA alternatives. We show that our DL-based tracker is able to match performance to the benchmarks in simple tracking tasks while outperforming the alternatives as the tasks become more challenging. These findings provide strong evidence of the applicability of DL also to the model-based setting, which we hope will foster further research in this direction.
|
|
| 9. |
- Pinto, Juliano, 1990, et al.
(författare)
-
Next Generation Multitarget Trackers: Random Finite Set Methods vs Transformer-based Deep Learning
- 2021
-
Ingår i: Proceedings of 2021 IEEE 24th International Conference on Information Fusion, FUSION 2021. ; , s. 1059-1066
-
Konferensbidrag (refereegranskat)abstract
- Multitarget Tracking (MTT) is the problem of tracking the states of an unknown number of objects using noisy measurements, with important applications to autonomous driving, surveillance, robotics, and others. In the model-based Bayesian setting, there are conjugate priors that enable us to express the multi-object posterior in closed form, which could theoretically provide Bayes-optimal estimates. However, the posterior involves a super-exponential growth of the number of hypotheses over time, forcing state-of-the-art methods to resort to approximations for remaining tractable, which can impact their performance in complex scenarios. Model-free methods based on deep-learning provide an attractive alternative, as they can, in principle, learn the optimal filter from data, but to the best of our knowledge were never compared to current state-of-the-art Bayesian filters, specially not in contexts where accurate models are available. In this paper, we propose a high-performing deeplearning method for MTT based on the Transformer architecture and compare it to two state-of-the-art Bayesian filters, in a setting where we assume the correct model is provided. Although this gives an edge to the model-based filters, it also allows us to generate unlimited training data. We show that the proposed model outperforms state-of-the-art Bayesian filters in complex scenarios, while matching their performance in simpler cases, which validates the applicability of deep-learning also in the model-based regime. The code for all our implementations is made available at https://github.com/JulianoLagana/MT3.
|
|
| 10. |
- Walton, Esther, et al.
(författare)
-
Brain Structure in Acutely Underweight and Partially Weight-Restored Individuals With Anorexia Nervosa : A Coordinated Analysis by the ENIGMA Eating Disorders Working Group
- 2022
-
Ingår i: Biological Psychiatry. - : Elsevier. - 0006-3223 .- 1873-2402. ; 92:9, s. 730-738
-
Tidskriftsartikel (refereegranskat)abstract
- BACKGROUND: The pattern of structural brain abnormalities in anorexia nervosa (AN) is still not well understood. While several studies report substantial deficits in gray matter volume and cortical thickness in acutely underweight patients, others find no differences, or even increases in patients compared with healthy control subjects. Recent weight regain before scanning may explain some of this heterogeneity. To clarify the extent, magnitude, and de-pendencies of gray matter changes in AN, we conducted a prospective, coordinated meta-analysis of multicenter neuroimaging data.METHODS: We analyzed T1-weighted structural magnetic resonance imaging scans assessed with standardized methods from 685 female patients with AN and 963 female healthy control subjects across 22 sites worldwide. In addition to a case-control comparison, we conducted a 3-group analysis comparing healthy control subjects with acutely underweight AN patients (n = 466) and partially weight-restored patients in treatment (n = 251).RESULTS: In AN, reductions in cortical thickness, subcortical volumes, and, to a lesser extent, cortical surface area were sizable (Cohen's d up to 0.95), widespread, and colocalized with hub regions. Highlighting the effects of un-dernutrition, these deficits were associated with lower body mass index in the AN sample and were less pronounced in partially weight-restored patients.CONCLUSIONS: The effect sizes observed for cortical thickness deficits in acute AN are the largest of any psychiatric disorder investigated in the ENIGMA (Enhancing Neuro Imaging Genetics through Meta Analysis) Consortium to date. These results confirm the importance of considering weight loss and renutrition in biomedical research on AN and underscore the importance of treatment engagement to prevent potentially long-lasting structural brain changes in this population.
|
|
