Imaging Informatics

Longitudinal DBT could help improve long-term cancer risk prediction

By Amerigo Allegretto

A longitudinal DBT model outperformed other models based on mammography and clinical factors in predicting five-year cancer risk.

August 13, 2026

AI tools may not meet breast radiologists' expectations in practice

By Amerigo Allegretto

AI tools in breast imaging may have more modest clinical impact on recall and biopsy rates.

August 12, 2026

Breast Mammo Computer 400 (3)

Model confidence, reader expertise affect reader-LLM collaboration

By Amerigo Allegretto

Successful human-LLM collaboration is associated with model confidence and reader expertise in chest imaging.

August 11, 2026

Example cases illustrate the double-edged role of large language model (LLM) rationale quality. (A) A 50-year-old female patient with a palpable anterior chest wall mass (correct diagnosis: solitary fibrous tumor of the pleura). The LLM incorrectly suggested synovial sarcoma but provided a persuasive rationale (quality score: 4 of 5), describing a well-defined nodule at axial contrast-enhanced CT spanning the pleura and intercostal space with heterogeneous strong enhancement. Four of five readers (80%) accepted the incorrect suggestion. FDG = fluorodeoxyglucose. (B) A 61-year-old male patient with diabetes presented with cough, sputum, and dyspnea (correct diagnosis: mucormycosis). The LLM correctly suggested mucormycosis with an excellent rationale (quality score: 5 of 5), integrating the reverse halo sign on serial axial CT images and rapidly progressive cavitary lesions. Three of five readers initially selected incorrect diagnoses; after LLM output, all five selected the correct diagnosis (100% acceptance). Blue and orange reader icons denote high- and low-expertise readers, respectively.

Many women interested in AI-based risk assessment for breast cancer

By Amerigo Allegretto

Survey findings also show that most women at higher risk are interested in additional screening tests.

August 10, 2026

Breast Cancer 400

DBT-based foundational model classifies breast density

By Amerigo Allegretto

DBT-DINO showed strong performance in classifying breast density from DBT images.

August 4, 2026

(A) Free-response receiver operating characteristic (FROC) curve for DBT-DINO and DINOv2 (Meta AI) on the detection task. DINOv2 achieved higher sensitivity than DBT-DINO across most false-positive thresholds. (B) Detection rates of DBT-DINO and DINOv2 for lesions that were ultimately found to be benign (n = 70) and lesions that were found to be cancerous (n = 66). Error bars represent 95% CIs estimated via patient-level bootstrapping with 1000 repetitions. Both models showed similar detection rates for cancerous lesions, whereas DINOv2 detected more benign lesions than DBT-DINO. (C) Example bounding box predictions for DBT-DINO and DINOv2 on the detection test set. Each image is a single two-dimensional section from a three-dimensional digital breast tomosynthesis volume; the first column shows a mediolateral oblique view, and the second and third columns show craniocaudal views. Both models struggled to detect very small lesions, such as the 3-mm-diameter lesion shown in the first column. FP = false positive.

Can LLMs fill the communication gap with referring clinicians?

LLMs could potentially help address multiple persistent pain points in radiology workflow.

August 4, 2026

MRI-based AI model maps local brain aging

By Amerigo Allegretto

The model showed differences in brain structures for normal cognition, mild cognitive impairment, and Alzheimer’s disease.

August 3, 2026

DeepHealth lands FDA nod for breast ultrasound AI software

By AuntMinnie.com staff writers

The software automates lesion detection, characterization, and reporting, according to the RadNet subsidiary.

July 31, 2026

Deephealth White On Purple

Podcast: The PACSMan Pontificates, Episode 4: The perfect PACS

By Michael J. Cannavo

In episode 4 of his The PACSMan Pontificates podcast series, Michael J. Cannavo shares his thoughts on the quest for the perfect PACS.

July 30, 2026

Pacs Man Pontification

Radiologists can identify three out of four AI-generated images

By Amerigo Allegretto

Radiologists found it particularly difficult to distinguish between real and AI-generated CT and MR images.

July 17, 2026

Top eight incorrectly classified synthetic images (Most convincing fake images). Images A, B, C, D, E, F, G and H were the most infrequently correctly classified images at 24%, 43%, 59%, 64%, 69%, 71%, 71% and 73% respectively. All images are AI-generated. Images A, B and C are head CTs. Images D and E are chest CTs. Image F is a cardiac CT. Image G is a musculoskeletal MRI. Image H is a head MRI. Images are republished under a Creative Commons license (CC BY 4.0).

Radiology dominates thirty years of FDA AI device approvals

The total annual authorization volume increased from a mean of 1.8 per year between 1995 and 2014 to 264 per year between 2023 and 2025.

July 16, 2026

Annual and cumulative FDA authorizations of AI/ML-enabled medical devices, 1995–2025.

Incorrect AI suggestions influence reader performance on mammography

By Amerigo Allegretto

The biggest impact on reader performance stemmed from false-negative AI suggestions.

July 14, 2026

A normal mammogram confirmed by three-year radiologic follow-up illustrates reader-marked regions of interest (ROIs) during (A) unaided (round 1) and (B) artificial intelligence (AI)–assisted (round 2) reading. Each colored dot represents an ROI for recall by a human reader. Readers could mark more than one ROI per case, represented by multiple dots of the same color. During AI-assisted reading, the AI system displayed three visible prompts: two with suspicion of malignancy scores of 35% (left mediolateral oblique [L MLO] and craniocaudal [L CC]) and one with a suspicion of malignancy score of 10% (right craniocaudal [R CC]), shown as polygonal overlays. Without AI, six of 10 readers (60%) marked a false-positive ROI. With AI assistance, this fell to two of 10 (20%). R MLO = right mediolateral oblique.

Page 1 of 1857