Supported imaging modalities
Chest X-Ray
Cardiomegaly, consolidation, pleural effusion, pneumothorax, hilar changes, interstitial patterns
Skeletal X-Ray
Fracture detection, joint space assessment, bone density observations, positional alignment
CT Scan Exports
Slice-level observations from CT image exports — abdominal, thoracic, neuro, musculoskeletal
MRI Images
Structural observations from MRI exports — brain, spine, MSK, abdominal, cardiac
Pathology Slides
Histological observations from slide images — cell morphology, tissue architecture, staining patterns
ECG Strips
Rhythm analysis, interval observations, ST segment changes, morphology patterns from image uploads
FAQ
Medical imaging AI — answered.
What types of medical images can ELLEXMED analyze?
EllexLens supports the following imaging modalities: plain X-rays (chest, skeletal, abdominal), CT scan image exports (JPEG/PNG), MRI image exports, pathology slide images, and ECG rhythm strips uploaded as images. DICOM format is not natively supported — images should be exported as JPEG or PNG from PACS systems for upload. Video formats are not supported.
Is EllexLens a diagnostic device or FDA-cleared medical device?
No. EllexLens is a clinical support tool — it provides AI-generated observations and potential findings as structured text for clinician review. It is not an autonomous diagnostic device and does not produce medical diagnoses. All outputs require clinician review, clinical correlation, and interpretation. ELLEXMED is aligned with the FDA Software as a Medical Device (SaMD) framework but EllexLens does not replace a radiologist or specialist.
What AI model powers EllexLens?
EllexLens is powered by Google Gemini 2.0 Flash in multimodal mode — a large multimodal model capable of analyzing both text and image inputs. The model generates structured observations based on visual patterns in the uploaded image combined with any clinical context provided (patient age, clinical question, relevant history).
How accurate is the imaging analysis?
AI imaging analysis accuracy varies significantly by modality, image quality, pathology complexity, and clinical context. Simple, high-contrast findings (e.g., cardiomegaly on CXR, bone fractures on X-ray) are recognized more reliably than subtle findings. EllexLens outputs should be treated as a second opinion or checklist aid — never as a primary diagnostic determination. False negatives and false positives are possible.
Are uploaded medical images stored?
Images are uploaded to secure cloud storage (encrypted at rest with AES-256) and associated with the patient's visit record in your organization's data scope. Images are not shared across organizations and are not used to train AI models. Your organization controls data retention and deletion.
Can EllexLens analyze images in the context of a specific clinical question?
Yes. When uploading an image, you can provide a clinical context prompt — e.g., 'Patient is a 65-year-old male, smoker, with 3-week productive cough. Review this CXR for consolidation or mass.' This context improves the relevance and specificity of the AI observations.