Fecal Classification

Why Fecal Classification

Built for clinical lab workflows

Designed for trained providers and laboratory staff — not for public self-diagnosis. Models provide triage-level signals while interpretation, documentation, and treatment decisions remain your responsibility.

Ensemble voting

Seven fine-tuned models vote on every slide. The majority decides — more robust than any single model.

3-stage pipeline

Fecal screening gates helminth detection, which gates species identification. Each stage narrows uncertainty.

Bounding-box overlays

Object detection draws boxes around each helminth species found, so you see exactly what the model sees.

Prediction history

Every result is stored with the original image. Review past predictions any time you sign back in.

Pipeline

Three stages, one clear path

Each stage narrows uncertainty before revealing detailed class predictions.

1

Stage 1

Fecal detection (ensemble)

7 models · majority vote

Seven fine-tuned TensorFlow models — VGG19, ResNet50, DenseNet169, EfficientNetB0, MobileNetV2, NASNetMobile, and ConvNeXtBase — each classify the slide independently. A majority-vote system combines their outputs: fecal or non-fecal. Non-fecal slides stop here.

2

Stage 2

Helminth screening

Binary classifier

Confirmed fecal samples pass to a dedicated binary classifier that distinguishes helminths from non-helminths. Non-helminth results are reported and the pipeline stops. Helminth-positive slides advance to species identification.

3

Stage 3

Species identification

11-class object detection

An object-detection model scans the slide for 11 parasitic helminth species — including Ascaris, Hookworm, Trichuris, and Taenia — drawing bounding boxes around each finding so clinicians see exactly where the model attended.

1

Stage 1

Fecal detection (ensemble)

7 models · majority vote

Seven fine-tuned TensorFlow models — VGG19, ResNet50, DenseNet169, EfficientNetB0, MobileNetV2, NASNetMobile, and ConvNeXtBase — each classify the slide independently. A majority-vote system combines their outputs: fecal or non-fecal. Non-fecal slides stop here.

2

Stage 2

Helminth screening

Binary classifier

Confirmed fecal samples pass to a dedicated binary classifier that distinguishes helminths from non-helminths. Non-helminth results are reported and the pipeline stops. Helminth-positive slides advance to species identification.

3

Stage 3

Species identification

11-class object detection

An object-detection model scans the slide for 11 parasitic helminth species — including Ascaris, Hookworm, Trichuris, and Taenia — drawing bounding boxes around each finding so clinicians see exactly where the model attended.

1

Stage 1 of 3

Fecal detection (ensemble)

Seven fine-tuned TensorFlow models — VGG19, ResNet50, DenseNet169, EfficientNetB0, MobileNetV2, NASNetMobile, and ConvNeXtBase — each classify the slide independently. A majority-vote system combines their outputs: fecal or non-fecal. Non-fecal slides stop here.

2

Stage 2 of 3

Helminth screening

Confirmed fecal samples pass to a dedicated binary classifier that distinguishes helminths from non-helminths. Non-helminth results are reported and the pipeline stops. Helminth-positive slides advance to species identification.

3

Stage 3 of 3

Species identification

An object-detection model scans the slide for 11 parasitic helminth species — including Ascaris, Hookworm, Trichuris, and Taenia — drawing bounding boxes around each finding so clinicians see exactly where the model attended.

Phase 1 ensemble

7 models, 1 consensus

Every uploaded slide is classified by seven independently fine-tuned TensorFlow / Keras architectures. Their outputs are combined through majority voting for a more reliable fecal vs non-fecal decision.

VGG19

Deep CNN· TensorFlow / Keras

ResNet50

Residual· TensorFlow / Keras

DenseNet169

Dense blocks· TensorFlow / Keras

EfficientNetB0

Compound scaling· TensorFlow / Keras

MobileNetV2

Lightweight· TensorFlow / Keras

NASNetMobile

NAS-optimized· TensorFlow / Keras

ConvNeXtBase

Modern CNN· TensorFlow / Keras

Majority vote

View all model details

Phase 3 detection

11 helminth species, localized on every slide

When helminths are confirmed, the object-detection model identifies and draws bounding boxes around these parasitic species directly on the microscopy image.

Ascaris lumbricoides

Giant roundworm — most common soil-transmitted helminth worldwide

Capillaria philippinensis

Intestinal capillariasis — causes chronic diarrhea and malabsorption

Enterobius vermicularis

Pinworm — the most common helminth in temperate climates

Fasciolopsis buski

Giant intestinal fluke — largest fluke infecting humans

Hookworm egg

Ancylostoma / Necator — leading cause of iron-deficiency anemia

Hymenolepis diminuta

Rat tapeworm — uncommon in humans, usually asymptomatic

Hymenolepis nana

Dwarf tapeworm — most common cestode in humans

Opisthorchis viverrine

Liver fluke — linked to cholangiocarcinoma risk

Paragonimus spp

Lung fluke — causes paragonimiasis, mimics tuberculosis

Taenia spp. egg

Tapeworm — beef (T. saginata) or pork (T. solium) tapeworm

Trichuris trichiura

Whipworm — infects the large intestine, common in tropics

Learn about these species

Access

Ready when you are

Predictions are available after you create an account and sign in. This landing page is informational — no uploads or inference run here.

Bring clinical clarity to every microscopic slide.

Create an account to unlock uploads and the full staged review workflow.

Get started free Sign in

AI-powered fecal screening for every microscopic slide.

Built for clinical lab workflows

Three stages, one clear path

7 models, 1 consensus

11 helminth species, localized on every slide

Ready when you are