Machine Learning Techniques for Automated Segmentation of Kidneys and Cysts in Autosomal Dominant Polycystic Kidney Disease: A Systematic Review
Bibliographic record
Abstract
Background: Autosomal Dominant Polycystic Kidney Disease (ADPKD) is the most common hereditary kidney disease. Total kidney volume (TKV) is used clinically as a prognostic biomarker, so its accurate determination is critical to determine treatment eligibility or inclusion into clinical studies. Recently, machine learning techniques have been applied to ADPKD to better segment kidneys or cysts. We performed a systematic review of machine learning models applied to ADPKD kidney and cyst segmentation. Methods: We conducted a literature search using relevant search terms including ADPKD, imaging, kidney or cyst segmentation, and machine learning techniques. The last search date was May 12, 2022. Studies were screened for relevance prior to inclusion in the systematic review. We identified seventeen studies. Two were excluded as they involved mouse models. We examined the patient and disease characteristics included in these studies, as well as the machine learning technique employed, and whether any external validation was performed. Results: Of the fifteen human studies eligible for inclusion (n=5243 kidney images[SJ1]), one used 3D ultrasound (n=66), three used CT images (n=502), and the remaining 11 used MR images (n=4675). The study sizes ranged from 11 up to 1445 patients, with 13 to 2400 scans. Four studies used axial cuts and 11 studies used coronal cuts to develop their model. The majority of models employed an artificial neural network approach; however, some studies utilized and compared the performance of different machine learning techniques. Only one study developed a model to segment kidney cysts. All models achieved high dice coefficient, sensitivity, specificity or recall when comparing with manual segmentation. Only one study performed external validation using a different cohort. Conclusions: Various machine learning techniques can accurately automate kidney segmentation in patients with ADPKD. Most studies relied on a small sample size and only one performed external validation. Whether these automated segmentation models can be deployed in a clinical environment or can outperform estimated TKV remains to be determined.
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 | 0.009 | 0.042 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.007 | 0.009 |
| Bibliometrics | 0.013 | 0.009 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.003 | 0.003 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.002 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.000 |
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".