Development of Whole Slide Imaging on Smartphones and Evaluation With ThinPrep Cytology Test Samples: Follow-Up Study
Bibliographic record
Abstract
BACKGROUND: The smartphone-based whole slide imaging (WSI) system represents a low-cost and effective alternative to automatic scanners for telepathology. In a previous study, the development of one such solution, named scalable whole slide imaging (sWSI), was presented and analyzed. A clinical evaluation of its iOS version with 100 frozen section samples verified the diagnosis-readiness of the produced virtual slides. OBJECTIVE: The first aim of this study was to delve into the quantifying issues encountered in the development of an Android version. It should also provide insights into future high-resolution real-time feedback medical imaging apps on Android and invoke the awareness of smartphone manufacturers for collaboration. The second aim of this study was to further verify the clinical value of sWSI with cytology samples. This type is different from the frozen section samples in that they require finer detail on the cellular level. METHODS: During sWSI development on Android, it was discovered that many models do not support uncompressed camera pixel data with sufficient resolution and full field of view. The proportion of models supporting the optimal format was estimated in a test on 200 mainstream Android models. Other factors, including slower processing speed and camera preview freezing, also led to inferior performance of sWSI on Android compared with the iOS version. The processing speed was mostly determined by the central processing unit frequency in theory, and the relationship was investigated in the 200-model simulation experiment with physical devices. The camera preview freezing was caused by the lag between triggering photo capture and resuming preview. In the clinical evaluation, 100 ThinPrep cytology test samples covering 6 diseases were scanned with sWSI and compared against the ground truth of optical microscopy. RESULTS: Among the tested Android models, only 3.0% (6/200) provided an optimal data format, meeting all criteria of quality and efficiency. The image-processing speed demonstrated a positive relationship with the central processing unit frequency but to a smaller degree than expected and was highly model-dependent. The virtual slides produced by sWSI on Android and iOS of ThinPrep cytology test samples achieved similar high quality. Using optical microscopy as the ground truth, pathologists made a correct diagnosis on 87.5% (175/200) of the cases with sWSI virtual slides. Depending on the sWSI version and the pathologist in charge, the kappa value varied between .70 and .82. All participating pathologists considered the quality of the sWSI virtual slides in the experiment to be adequate for routine usage. CONCLUSIONS: Limited by hardware and operating system support, the performance of sWSI on mainstream Android smartphones did not fully match the iOS version. However, in practice, this difference was not significant, and both were adequate for digitizing most of the sample types for telepathology consultation.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.010 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".