Active-DiNTS: Active Differentiable Network Topology Search
Authors: Gean Trindade Pereira, Thierry Urruty, Muriel Visani, André C. P. L. F. de Carvalho
Organizations: Institute of Mathematics and Computer Sciences (ICMC) University of São Paulo São Carlos, 13560-970, São Paulo, Brazil · XLIM, Université de Poitiers Université de Limoges, CNRS Poitiers, 86000, Vienne, France · Laboratoire Informatique, Image et Interaction (L3i) La Rochelle Université La Rochelle, 17042, Charente-Maritime, France
Neural Architecture Search (NAS) has proved to be a strong alternative to manual network design, but applying it to 3D medical image segmentation is limited by two well-known costs, large annotation budgets and multi-GPU clusters. Thus, this paper introduces Active-DiNTS (Active Differentiable Network Topology Search), an approach that embeds pool-based Active Learning (AL) into a bi-level differentiable topology search to perform architecture discovery and label curation jointly. At each query round, unlabeled MRI volumes are ranked by one of three uncertainty signals (Entropy, Variance, or Standard Deviation), and only the top-ranked volumes are sent to an oracle for annotation. The new labels feed two interlocked stages. Network weights are updated in an outer loop, while the macro/micro topology of a U-Net-style backbone is refined in an inner loop. Three AL regimes (weights-only, topology-only, joint) expose the speed-accuracy trade-off. Evaluations on the Medical Segmentation Decathlon (MSD) Task01 BrainTumour benchmark showed that Active-DiNTS surpasses DiNTS, C2FNAS, and nnU-Net in Dice, with gains of about 10 percentage points on Edema and 5 points on Non-Enhancing core, on a single GPU and using a fraction of the labeled volumes. The discovered architectures are denser and more FLOP-heavy than prior baselines, but remain competitive in trainable parameters and peak memory; the fastest search variant finishes in under 0.25 GPU-days, over 27x faster than the eight-GPU DiNTS search. Together, these results indicate that pairing differentiable NAS with active data acquisition is a practical recipe for accurate 3D segmentation under realistic constraints.
Figures & tables
Figure 1: Topology and cell search spaces explored by Active-DiNTS. The macro graph (top) interconnects L=12 layers across D=4 resolutions; green dashed edges carry one of O=5 cell operations. Adapted from He et al. [ 9 ] .
Figure 2: Active-DiNTS workflow. (1) Unlabeled pools UW and UT are scored by uncertainty; (2) the oracle annotates the most informative volumes; (3) bi-level optimization alternates between weights and topology.
Figure 3: Learning curves for the joint AL setup. Uncertainty-driven strategies (Entropy, Variance, Std) are contrasted with Random selection and the DiNTS baseline (horizontal line).
Figure 4: Dice curves averaged over uncertainty heuristics for each AL setup (weights-only, topology-only, joint), shown against the Random and DiNTS baselines.
Method
Most frequent op
Least frequent op
Total ops
DiNTS
3D conv 3×3×3 (34.0%)
P3D 3×1×3 (12.4%)
97/120
Active-DiNTS (Random)
3D conv 3×3×3 (26.7%)
skip (15.0%)
120/120
Active-DiNTS (Entropy)
3D conv 3×3×3 (27.5%)
P3D 1×3×3 (15.0%)
120/120
Active-DiNTS (Variance)
3D conv 3×3×3 (29.2%)
P3D 3×3×1 (13.3%)
120/120
Active-DiNTS (Std)
3D conv 3×3×3 (26.6%)
skip (15.8%)
120/120
Table 1: Operation statistics for the best Active-DiNTS architectures versus DiNTS.
In practical settings, medical image segmentation models are often developed with limited annotated data rather than fully labeled datasets. Training frequently begins in ultra-low labeled regimes where only a small number of volumes are annotated. In such scenarios, practitioners must simultaneously decide which cases to annotate and how to best use the remaining unlabeled data. Although active learning (AL) and semi-supervised learning (SSL) both target annotation scarcity, they are typically designed and optimized independently, resulting in objective mismatch and unstable training during early-stage "cold start" conditions. We propose RegAL, a unified active semi-supervised framework governed by a shared topology-aware Pareto optimization that couples sample acquisition with unlabeled data utilization. RegAL evaluates images along three complementary axes, voxel-wise uncertainty, feature diversity, and a novel topological consistency metric, to select anatomically informative edge cases for annotation. On the other hand, the same criteria are used to identify geometrically stable atlas candidates for diffeomorphic registration-guided augmentation to train a self-supervised Mean Teacher segmentation network. Across BraTS 2021, dHCP, and ProstateX, RegAL remains stable with few labeled volumes and consistently outperforms state-of-the-art AL, SSL, and active semi-supervised baselines across Dice and boundary-distance (ASD, HD95) metrics under extreme annotation scarcity.
Bahram Jafrasteh, Cheng Wan, Heejong Kim +2
Department of Radiology, Weill Cornell Medicine, New York, NY, USA · Department of Electrical and Computer Engineering, Cornell University, Ithaca, NY, USA
Deep learning for 3D medical image segmentation requires extensive manual annotations, a major bottleneck in volumetric medical imaging. Active learning aims to reduce this burden by selecting informative samples for annotation, but most methods assume that an initial labeled set is already available. This leaves the cold-start problem largely unresolved: how to select the first volumes from a fully unlabeled pool before any task-specific model is trained. We propose CSCS, a Curriculum-Stratified Cold-Start framework that adapts initial sample selection to the structure of the unlabeled dataset. CSCS combines two self-supervised, label-free signals: local typicality, measuring representativeness in the embedding space, and reconstruction-based uncertainty, used as a proxy for sample difficulty. These signals are combined through a weighted geometric score, where the weighting is determined by a closed-form pacing rule based on the effective annotation budget and the Difficulty-Coverage Ratio, a pool-level statistic measuring the alignment between difficulty and representativeness. We evaluate CSCS on four 3D medical image segmentation benchmarks: BraTS, FeTA, Spleen, and an in-house fetal MRI dataset. Using nnU-Net as downstream segmentation model, CSCS shows consistently competitive performance across datasets and annotation budgets, with the strongest gains in low-to-mid annotation regimes. These results suggest that dataset-aware cold-start initialization can improve the robustness of active learning for 3D medical image segmentation by adapting sample selection to the geometry of the unlabeled pool.
This paper proposes a novel medical image segmentation framework, MNAS-Unet, which combines Monte Carlo Tree Search (MCTS) and Neural Architecture Search (NAS). MNAS-Unet dynamically explores promising network architectures through MCTS, significantly enhancing the efficiency and accuracy of architecture search. It also optimizes the DownSC and UpSC unit structures, enabling fast and precise model adjustments. Experimental results demonstrate that MNAS-Unet outperforms NAS-Unet and other state-of-the-art models in segmentation accuracy on several medical image datasets, including PROMISE12, Ultrasound Nerve, and CHAOS. Furthermore, compared with NAS-Unet, MNAS-Unet reduces the architecture search budget by 54% (early stopping at 139 epochs versus 300 epochs under the same search setting), while achieving a lightweight model with only 0.6M parameters and lower GPU memory consumption, which further improves its practical applicability. These results suggest that MNAS-Unet can improve search efficiency while maintaining competitive segmentation accuracy under practical resource constraints.