Digital Pathology Platform

9DA Solutions

One platform for oral cancer care

Detect Oral Cancer and Recurrence Reliably

Systemically & Reliably

The 9DA platform stands for deep learning-supported diagnostic tool – all in one system. We offer an online system to detect oral cancer and to predict recurrence. Physicians simply upload their whole-slide scanned pathology images onto the platform to detect the presence of oral cancer. To get an overview of patient post-treatment status, selecting regions of interest on whole-slide images can provide insights into prognosis.

Direction for use

  1. Upload and manage whole-slide brightfield pathology images
  2. Annotate and share WSIs across clinical teams
  3. Secure cloud workflow for centralized asset management

Key benefits

  • Streamlines digital pathology operations across sites
  • Enables remote collaboration for distributed teams
  • Foundation for 9DA AI diagnostic modules (LNTD & MILP)

Detect the Presence of Oral Cancer

Simpler. Faster.

Our diagnostic tool, LNTD, classifies patients into both cancer and non-cancer cases, thereby relieving the significant burden and stress on healthcare systems. By reducing the number of unnecessary referrals, biopsies, and specialist consultations for benign conditions, the tool allows medical professionals to focus their limited time and resources on patients who need urgent cancer care. Ultimately, LNTD serves not only as a digital tool but also as a support mechanism for overwhelmed healthcare systems, especially in regions with limited access to oncology specialists.

Direction for use

  1. For analysis on haematoxylin and eosin-stained pathology slides
  2. Support whole slide-scanned brightfield images at 40× magnification
  3. Mainly support ndpi, svs, and tiff formats

Key benefits

  • Distinguishing malignant tumor from non-tumor cases
  • Segmentation maps are visible and clear for analysis
  • Report for oral cancer detection is professionally customized

Predict Oral Cancer Recurrence

Efficient. Revolutionary. Explainable.

To assess patient status after treatment, MILP accepts brightfield and/or confocal hematoxylin and eosin-stained pathology images for recurrence analysis. The model classifies each case as early recurrence or late/no recurrence, then uses LIME (Local Interpretable Model-agnostic Explanations) to highlight the slide regions that most influenced that prediction. Results include a logistic value and a probability for clinical review.

Direction for use

  1. For analysis on haematoxylin and eosin-stained pathology slides
  2. Support whole slide-scanned brightfield at 40× magnification and confocal images at 10× magnification
  3. Mainly support .ndpi, .svs, and .tiff formats for brightfield images
  4. Only support .czi format for confocal images
  5. Review LIME segment highlights showing which regions drove each recurrence prediction

Key benefits

  • Distinguishing early-recurrence (recurrence within 5 years) from late/no-recurrence (recurrence beyond 5 years or no recurrence) cases
  • After each prediction, LIME highlights the slide regions that most influenced the model's classification
  • Results include logistic values and probabilities so clinicians can review why MILP reached each outcome
  • Report for recurrence prediction is professionally customized