Search Results

InferCare RECIST is a post-processing software application used to display, process, analyze, quantify and manipulate multi-time-point CT images. It is intended to be used by trained medical professionals in evaluating and managing tumors in various organs, tissues, and other anatomical structures based on RECIST criteria.

InferCare RECIST offers functionalities for lesion measurement, registration, tracking, and RECIST report. It provides tools for interactive segmentation, and 3D reconstruction visualization.

The software utilizes artificial intelligence algorithms for automated lesion segmentation and registration, and the results require confirmation by a medical professional. The software's artificial intelligence algorithms are intended for patients aged 21 years and older.

InferCare RECIST is based on advanced image processing technologies and temporal sequence analysis methods. The software can be installed in a healthcare facility or a cloud-based platform. It ingests medical imaging data compliant with the DICOM standard and supports multi-timepoint image registration. It allows automatic navigation to match lesion locations across different image series based on physician-selected lesions, ensuring accurate tracking and comparison of the same lesion across follow-up studies.

InferCare RECIST integrates machine learning based algorithms to provide interactive segmentation tools, enabling users to generate 3D segmentation and measurement results simply by clicking on the target lesion.

In addition, the system offers a variety of image processing tools, such as window width/level adjustment, multi-planar reconstruction (MPR), and maximum intensity projection (MIP), to meet the needs of different clinical viewing scenarios. Through the integrated use of these technologies, InferCare RECIST delivers accurate lesion measurements, follow-up trend analyses, and RECIST-standard evaluation reports, effectively supporting physicians in making precise diagnostic and treatment decisions.

Here's a breakdown of the acceptance criteria and study details for the InferCare RECIST device, based on the provided FDA 510(k) clearance letter:

Acceptance Criteria and Reported Device Performance

Note: For the lesion segmentation accuracy and measurement, the document states "Segmentation accuracy was set as the primary endpoint, and Long/Short diameter measurement was set as the secondary endpoint." While explicit numerical acceptance criteria are not provided in the text, the reported values are presented as meeting the performance test requirements by implication of the submission being cleared and the statement that "standalone performance testing has demonstrated that the subject device meets the predetermined target values."

Study Details for Device Performance Evaluation

Here's a breakdown of how the device performance was proven:

1. Sample sizes used for the test set and the data provenance:

• Image Registration Performance Test:
  • Test Set Size: "a total of 212 target objects including 184 paired, 21 new and 7 disappeared target objects were annotated as ground truth." This was derived from "86 patient-scans".
  • Data Provenance: Not explicitly stated (e.g., country of origin). It is retrospective as it involves annotated existing data.
• Lesion Segmentation Performance Test:
  • Test Set Size: "a total of 102 cases, including 42 cases of lung nodules, 23 cases of liver lesions, 23 cases of kidney lesions, and 14 cases of lymph node lesions."
  • Data Provenance: Not explicitly stated (e.g., country of origin). It is retrospective as it involves annotated existing data.

2. Number of experts used to establish the ground truth for the test set and the qualifications of those experts:

• This information is not provided in the given document. The document only states that "212 target objects... were annotated as ground truth" and refers to "ground truth" for segmentation, but does not specify who performed the annotations or their qualifications.

3. Adjudication method for the test set:

• This information is not provided in the given document. It is unknown if multiple experts annotated and how disagreements were resolved.

4. If a multi reader multi case (MRMC) comparative effectiveness study was done, if so, what was the effect size of how much human readers improve with AI vs without AI assistance:

• No, an MRMC comparative effectiveness study was not explicitly done or reported in this document. The performance data section focuses solely on the AI algorithm's standalone performance ("Performance of AI algorithm" and "standalone performance testing"). The Indications for Use section states the AI results "require confirmation by a medical professional," implying a human-in-the-loop workflow, but a specific study measuring human improvement with AI assistance is not detailed.

5. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done:

• Yes, a standalone algorithm-only performance study was conducted. The sections "Performance of AI algorithm" and the statement "standalone performance testing has demonstrated that the subject device meets the predetermined target values" explicitly indicate this.

6. The type of ground truth used (expert consensus, pathology, outcomes data, etc.):

• For Image Registration, the ground truth involved "annotated" target objects (paired, new, disappeared). This strongly implies expert annotation of lesion presence and correspondence across time points.
• For Lesion Segmentation, the ground truth was used to compare against the algorithm's segmentation and measurements (Dice score, Long/Short diameter MAE). This also strongly implies expert annotation/segmentation as the reference.
• Pathology or outcomes data are not mentioned as the direct ground truth for these specific performance metrics.

7. The sample size for the training set:

• This information is not provided in the given document. The document only describes the test set and performance results.

8. How the ground truth for the training set was established:

• This information is not provided in the given document. Since the training set size and characteristics are not mentioned, neither is the method for establishing its ground truth.

Ask a specific question about this device