---
title: 'MIDOG 2025 Task 2: Atypical Mitosis Classification'
url: https://www.emergentmind.com/topics/midog-2025-task-2
type: topic
---

# MIDOG 2025 Task 2: Atypical Mitosis Classification

Searching arXiv for MIDOG 2025 Task 2 papers to ground the article in the latest challenge literature.
MIDOG 2025 Task 2 is the atypical mitotic figure classification track of the Mitosis Domain Generalization challenge, formulated as a binary classification problem on image patches centered on confirmed mitotic figures. The task requires assigning each patch to either a normal mitotic figure (NMF) or an atypical mitotic figure (AMF) under substantial domain shift arising from differences in tumor type, species, laboratory, staining protocol, and scanner. Across the 2025 literature, Task 2 is consistently framed as a test of domain-robust fine-grained morphology recognition, with balanced accuracy as the principal metric because AMFs are a minority class and because cross-domain robustness, rather than raw accuracy, is the organizing challenge [2508.18831].

## 1. Definition and challenge scope

MIDOG 2025 comprises two tracks: robust mitotic figure detection and atypical mitotic figure classification. Task 2 corresponds to the latter and is defined as follows: given an image patch centered on a mitotic figure, classify the mitosis as typical or atypical, equivalently normal or atypical, under strong variation in scanners, centers, tissues, tumor types, and species [2508.21035]. The input format is patch-level rather than whole-slide level, and the output is binary, typically encoded as class 0 for NMF and class 1 for AMF, or equivalently typical versus atypical [2508.18831].

The clinical motivation stated in the challenge literature is that atypical mitoses reflect abnormal division patterns and are associated with tumor aggressiveness, genomic instability, and prognostic relevance. The computational difficulty is not the presence of a mitotic figure per se, since Track 2 patches are centered on confirmed mitoses, but the discrimination of subtle morphological irregularities from the range of appearances exhibited by normal mitotic phases [2508.21035]. This is further complicated by the fact that both normal and atypical mitoses display high intra-class variability, while their inter-class differences are often visually subtle [2508.18831].

The domain generalization aspect is explicit. The challenge evaluates performance on unseen domains, and the domain concept itself is multi-factorial: scanner, laboratory or origin, species, and tumor type are recurrently treated as components of domain identity [2508.20745]. A plausible implication is that Task 2 is best understood not merely as binary classification, but as binary classification under structured distribution shift.

## 2. Data resources, annotations, and label structure

The official MIDOG 2025 atypical training data are described in several closely related but not identical ways across submissions. One formulation reports 11,939 mitotic figures from the MIDOG 2025 Atypical Training Set, with 10,191 normal mitotic figures and 1,748 atypical mitotic figures, derived from 454 labeled images and nine distinct domains, with a test set covering 120 cases and 12 distinct tumor types including both human and veterinary pathology [2508.18831]. Another describes the MIDOG 2025 Atypical Training Set as encompassing 11,939 mitotic figures from all seven domains of MIDOG++ [2508.21035]. A third, post-challenge synthesis states that for MIDOG 2025 all 11,939 mitotic figures in MIDOG++ were re-annotated as AMF versus NMF and released as a public training dataset, while the final test set comprised hotspot-region mitotic figures from 122 whole-slide images across 12 tumor domains [2606.07368]. The differing domain counts in the challenge papers reflect differences in how domain partitions were described in individual submissions rather than a single harmonized convention.

The available public ecosystem extends beyond the official training set. Frequently reused external resources include AMi-Br, OMG-Octo Atypical, AtNorM-Br, LUNG-MITO, and Jahanifar’s mitosis subtyping dataset [2509.02591], [2509.02595], [2509.02601]. These additions are used to enlarge morphological coverage and domain diversity. For example, AMi-Br is repeatedly described as containing 3,720 mitotic figures, of which 832 are atypical and 2,888 normal [2508.21035]. OMG-Octo Atypical is explicitly described as a refinement of the original OMG-Octo database to incorporate atypical mitosis labels [2509.02585].

The annotation structure is also important. In the official training set, labels are image-level binary subtype labels per mitotic figure crop [2508.18831]. Some works further exploit multi-expert annotations. One study converts three pathologist annotations into soft labels
\[
\tilde{y}_i = \frac{1}{3}\sum_{k=1}^3 y_i^{(k)},
\]
so that expert disagreement is represented probabilistically rather than collapsed immediately to a hard class [2509.02601]. Another paper formalizes a separate notion of instance difficulty by designating patches with full expert agreement as easy and those with disagreement as hard, producing a secondary hardness label in addition to phenotype [2509.10502]. This suggests that label uncertainty itself is treated as a meaningful signal in some Task 2 formulations.

## 3. Morphological and statistical difficulties

Three recurrent properties define the difficulty of MIDOG 2025 Task 2. First is class imbalance. In the official training set, AMFs are a minority; one paper characterizes them as approximately 20% of all mitotic figures, while also providing the explicit ratio
\[
\text{AMF proportion} \approx \frac{1748}{10191 + 1748} \approx 0.146
\]
and noting that “\(\sim 20\%\)” is a rounded characterization [2508.18831]. Another combined dataset after deduplication reports 1,771 AMFs and 10,168 NMFs, a class ratio of approximately \(1:5.74\) [2509.02601]. The challenge overview emphasizes that AMFs remain a minority in nearly all domains of the final test set [2606.07368].

Second is morphological ambiguity. Atypical mitoses are described as cells that do not follow a structured division pattern and may exhibit tripolar or multipolar spindles, asymmetrical chromatin distribution, lagging chromosomes, ring-shaped mitoses, or fragmented chromosome configurations [2508.21035], [2606.07368]. Normal mitoses encompass the canonical phases of prophase, metaphase, anaphase, and telophase [2508.21035]. The challenge papers repeatedly note that the distinction between NMF and AMF can be subtle even for experts, and that inter-observer variability is substantial [2509.02640], [2508.21041].

Third is domain heterogeneity. Differences in scanner hardware, acquisition pipeline, stain appearance, tissue morphology, and species affect color distribution, texture, background microenvironment, and artefact patterns [2508.21035]. Post-challenge analysis found significant variation in Track 2 performance across tumor domains, with human astrocytoma among the easiest and feline GI lymphoma among the hardest, indicating persistent domain-specific blind spots even in top-performing models [2606.07368].

## 4. Predominant methodological families

Task 2 methods fall into a small number of clearly defined technical families.

The first family is modern convolutional classification. ConvNeXt is especially prominent. One submission used ConvNeXt V2 Base with a single-neuron output head, BCEWithLogitsLoss, 60% center cropping, five-fold stratified cross-validation, and fold averaging, achieving a cross-validation balanced accuracy of \(0.8314 \pm 0.0261\) and development-phase balanced accuracy of 0.8831 [2508.18831]. Another large comparative study found ConvNeXt-large to be the best single model among ViT, ResNet, DenseNet, ConvNeXt, and EfficientNet variants, with cross-validation balanced accuracy \(0.8794 \pm 0.0082\), and used a ConvNeXt ensemble to achieve preliminary balanced accuracy 0.86 [2509.02597]. A further ConvNeXt Small–based study trained on AMi-Br, AtNorM-Br, MIDOG++, and OMG-Octo, combining weighted sampling with extensive histopathology-specific augmentation, and reported preliminary leaderboard balanced accuracy 0.8961 and ROC AUC 0.9561 [2509.02595].

The second family is pathology foundation models with parameter-efficient adaptation. Virchow, Virchow2, UNI, UNI2-h, H-optimus-0, and HIBOU all appear in the 2025 literature. LoRA is the dominant PEFT mechanism. Ramchandani et al. evaluated Virchow, Virchow2, and UNI with LoRA and found their best approach, Virchow with rank-8 LoRA and three-fold ensembling, reached balanced accuracy 88.37% on the preliminary test set [2509.16935]. A DINOv3-H+ model pretrained on natural images and adapted with LoRA on query and value projections, using only about 650k trainable parameters, achieved preliminary balanced accuracy 0.8871 [2508.21041]. Another foundation-model submission used H-optimus-0 with LoRA, soft labels, adaptive focal loss, hard negative mining, supervised contrastive regularization, and a domain-adversarial head, achieving mean leave-one-domain-out balanced accuracy \(0.851 \pm 0.037\) across 10 domains [2509.02601].

The third family is prompt-based or other parameter-efficient adaptation of pathology transformers. Meng et al. found that visual prompt tuning on UNI2-h improved over a LoRA baseline, increasing balanced accuracy from 0.8305 to 0.8711, and that adding TTA with Vahadane and Macenko stain normalization further raised it to 0.8837 with ROC AUC 0.9513 [2509.02640].

The fourth family is multi-task or object-centric learning. Percannella et al. proposed a multi-task network with a shared Pre-Activated ResNet-50 backbone and three heads: image-level classification, binary mitosis segmentation, and pixel-level mitosis classification. Under leave-one-domain-out evaluation, the multi-task model improved balanced accuracy over a single-task baseline on validation, unseen test domains, and external AMi-Br, and achieved 0.856 on the preliminary MIDOG25 test set [2508.21035]. A separate teacher–student framework coupled UNet-based segmentation, contrastive learning, domain-adversarial training, and a multi-scale classifier head, reaching 0.8418 balanced accuracy on the preliminary test set for Track 2 [2509.03614].

The fifth family emphasizes domain generalization through explicit training-time regularization. “Mix, Align, Distil” combines MixStyle in early and mid DenseNet-121 stages, CBAM-based cross-domain alignment using weak metadata, and EMA-teacher distillation with temperature-scaled KL divergence. Its preliminary leaderboard results were balanced accuracy 0.8762, sensitivity 0.8873, specificity 0.8651, and ROC AUC 0.9499 [2508.20745].

## 5. Training strategies, imbalance handling, and evaluation protocol

Despite substantial architectural diversity, several training practices recur.

Balanced accuracy is the dominant primary metric. It is typically defined as
\[
BA = \frac{TPR + TNR}{2},
\]
with \(TPR = \frac{TP}{TP+FN}\) and \(TNR = \frac{TN}{TN+FP}\) [2508.18831], [2508.21035]. Challenge-wide reporting also includes ROC AUC, and many papers additionally report sensitivity and specificity [2508.18831], [2509.16935], [2509.02640].

Cross-validation protocols vary. Stratified five-fold cross-validation is common in pure classifier pipelines [2508.18831], [2509.02595], [2509.10502]. Leave-one-domain-out is used when domain shift itself is the principal experimental variable [2508.21035], [2509.02601]. Some submissions combine the official MIDOG training set with external datasets and perform four-fold cross-validation on the merged corpus [2509.02597], [2508.21041].

Loss design reflects class imbalance. BCEWithLogitsLoss is frequently used without explicit class weights in ConvNeXt pipelines [2508.18831], [2509.02597]. Other methods explicitly address imbalance through focal loss [2508.21041], [2509.10502], adaptive focal loss with dynamic positive weighting [2509.02601], or weighted binary cross-entropy based on inverse class frequency [2508.21035]. Sampling-based corrections are also widespread: WeightedRandomSampler is used both for balancing classes [2509.16935], [2509.02595] and for weighting datasets or domains [2509.02591], [2509.02601].

Augmentation is central to domain robustness. Geometric transforms such as flips, rotations, random crops, shift-scale-rotate, elastic transforms, grid distortion, and perspective warping are common [2508.18831], [2509.02595], [2508.20745]. Histopathology-specific color processing includes stain augmentation, ColorJitter, hue-saturation perturbations, CLAHE, defocus blur, Gaussian noise, ISO noise, and channel perturbations [2509.02595], [2508.21041]. Several methods incorporate stain normalization at training or test time using Macenko or Vahadane [2509.02640]. Fourier Domain Adaptation and fisheye transformations are also used in at least one pathology foundation model ensemble [2509.02591].

One especially influential preprocessing choice is center cropping. In the ConvNeXt V2 submission, a 60% center crop reduced a \(128 \times 128\) patch to approximately \(77 \times 77\) before resizing back to 128. This improved training stability and raised balanced accuracy by \(+0.0402\) relative to a no-cropping condition in converged folds, largely through a sensitivity gain of \(+0.0867\) for AMFs [2508.18831].

## 6. Performance patterns and post-challenge synthesis

The post-challenge organizer analysis provides the clearest global picture. MIDOG 2025 Track 2 received 21 submissions, and balanced accuracy on the final test set reached up to 0.908 [2606.07368]. The winning method was a DINOv3-H model with LoRA adaptation trained on MIDOG 2025, AMi-Br, and OMG-Octo Atypical, obtaining balanced accuracy 0.908 without ensembling, while the strongest ROC AUC, 0.971, was reported by a ConvNeXt v2–based ensemble [2606.07368]. The official baseline, an EfficientNetV2 classifier, achieved balanced accuracy 0.827 [2606.07368].

A notable result of the organizer ablation is that ensembling improves Track 2 performance on average by approximately 1.3 percentage points in balanced accuracy, whereas TTA shows no relevant improvement, with only about 0.3 percentage points mean gain [2606.07368]. This is consistent with several individual papers in which ensemble averaging or majority voting is treated as a major robustness mechanism [2508.18831], [2508.21035], [2509.02597], [2509.02591].

Performance is not uniform across tumor types. Organizers report significant differences across the 12 tumor domains, with “blind spots” in rare or highly pleomorphic malignancies [2606.07368]. Human astrocytoma shows the highest median balanced accuracy, while feline GI lymphoma shows the lowest, with reduced AMF recall [2606.07368]. This suggests that cross-domain generalization in Task 2 remains limited by the representational coverage of atypical morphologies rather than by a single universal failure mode.

At the individual-method level, strong preliminary performances cluster in the high-0.87 to low-0.89 balanced-accuracy range for many competitive systems: 0.8837 for Virchow plus LoRA [2509.16935], 0.8837 for VPT plus UNI2-h with stain-normalization TTA [2509.02640], 0.8871 for DINOv3-H+ with LoRA [2508.21041], 0.8961 for ConvNeXt Small with histopathology-specific augmentation [2509.02595], and 0.9107 in one data-centric pan-cancer submission that aggregated five datasets and emphasized stain-aware augmentation and model selection around ConvNeXt-tiny [2509.02585]. These figures indicate that the field converged quickly on a relatively narrow performance band, with differences often traceable to dataset breadth, augmentation policy, and adaptation mechanism rather than to a single dominant backbone class.

## 7. Limitations, controversies, and future directions

The challenge literature repeatedly identifies three unresolved issues.

The first is incomplete domain generalization. Even top-performing models exhibit marked domain-specific variability [2606.07368]. Some methods explicitly acknowledge that their gains under domain shift are modest, for example the multi-task network that improves balanced accuracy by approximately 0.007 on unseen domains relative to a single-task baseline [2508.21035]. A plausible implication is that many current methods reduce but do not eliminate dependence on stain, scanner, or tissue-context cues.

The second is interpretability. Several papers note that these systems remain black-box classifiers with limited interpretability regarding the morphological features driving AMF versus NMF decisions [2508.18831]. This matters because AMF classification is clinically consequential and already subject to expert disagreement. Some future directions proposed in the literature include Grad-CAM, saliency methods, or concept-based analysis, but these are generally presented as prospective additions rather than evaluated components [2508.18831], [2509.02601].

The third is the status of atypical morphology itself as a label space. Most submissions treat Task 2 as binary subtype classification, but the morphological guide underlying the annotations distinguishes multiple abnormal patterns such as multipolarity, lagging chromosomes, and anaphase bridges [2606.07368]. Several papers suggest, implicitly or explicitly, that finer-grained AMF subclassification could be a future extension [2509.02585], [2606.07368]. This suggests that the current binary formulation is both practical and reductive: practical because it provides a tractable benchmark, reductive because it compresses morphologically diverse abnormalities into a single positive class.

A further methodological tension concerns whether task-specific, lightweight CNNs or very large foundation models are preferable. Some submissions report that compact ConvNeXt variants or even ConvNeXt-tiny trained from scratch outperform more elaborate foundation-model pipelines when sufficient multi-domain data are assembled [2509.02585]. Others show clear benefits from Virchow, UNI, or DINOv3 with LoRA or VPT [2509.16935], [2509.02640], [2508.21041]. The challenge-level results indicate that both paradigms are competitive. This suggests that success in MIDOG 2025 Task 2 depends at least as much on data composition, augmentation, and calibration as on the choice between CNN and transformer backbones.

Overall, MIDOG 2025 Task 2 established atypical mitotic figure classification as a distinct, reproducible benchmark in computational pathology, moving beyond mitosis detection toward morphology-aware prognostic phenotyping. The challenge demonstrated that balanced accuracies near 0.91 are achievable on a multi-domain benchmark, yet also showed that rare tumor types, pleomorphic domains, and annotation ambiguity remain major constraints on generalization [2606.07368].

Source: https://www.emergentmind.com/topics/midog-2025-task-2