Search Results

The Leica FL800 is a surgical microscope accessory used in viewing intra-operative blood flow in the cerebral vascular area.

The Leica FL800 device is a surgical microscope accessory used for viewing intra-operative blood flow in the cerebral vascular area. It allows the surgical microscope to produce excitation light and resolve florescence light from the fluorescent agent ICG. The Leica FL800 is an accessory to the Leica M520 OH3 Neurosurgical Microscope and is comprised of five components: Standard Leica M520 OH3 Microscope, Sony XC-E150 NIR (near infra-red) camera, Leica Dual CCD Surgical Microscope Camera Adapter, Leica modification of standard 300 Watt Xe light source, and Leica FL800 electronic control unit. It is labeled for use with Akorn Inc.'s Indocyanine-Green (ICG).

Here's an analysis of the Leica FL800's acceptance criteria and the study used to demonstrate its performance, based on the provided document:

Acceptance Criteria and Device Performance

The provided document does not explicitly state specific quantitative acceptance criteria for metrics like sensitivity, specificity, accuracy, or a defined threshold for "excellent quality" beyond the broad statement that "NIR signal to camera must exceed 40% of available NIR light."

However, based on the conclusion of the clinical validation, the de facto acceptance criterion appears to be:

• Ability to successfully visualize blood flow.
• Confirmation of surgical outcomes (patency of bypasses, optimal aneurysm clipping) aligning with subsequent sonography/postoperative angiography and surgeon's intraoperative decision-making.
• Performance (safety and effectiveness) comparable to the predicate device (Novadaq SPY).

Study Details

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

   • Test Set Sample Size: 18 patients.
     • 7 cases were performed at 20% NIR (signal too weak)
     • 11 cases were performed at 40% NIR (all excellent results)
   • Data Provenance: Prospective clinical test conducted from January to April 2006 at the University of Illinois – Chicago (USA) under an IRB. A "validation of concept" was also performed in September 2005 in Dresden, Germany, but this was more for initial issue identification and hardware refinement rather than formal outcome assessment of the final device. Specific case results from the Dresden validation are not reported in detail for performance metrics.

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

   • The document implies that the neurosurgeon performing the procedure was the primary "expert" in the loop, making real-time decisions based on the FL800's output. The term "experts" in the context of independent ground truth adjudication is not used.
   • The study involved clinical procedures performed by surgeons; however, the ground truth was confirmed by comparison with sonogram and postoperative angiography results, which are diagnostic imaging modalities typically interpreted by radiologists or interventionalists. The number of such specialists and their specific qualifications are not specified in the document.

3. Adjudication method for the test set:

   • The document states that for bypass cases, "the results were followed up with a sonogram that confirmed that the qualitative ICG fluorescence was correct in terms of the sonogram." Additionally, "Each patient was followed up with a postoperative angiography."
   • For aneurysm cases, the device "clearly showed that the clipping was not optimal and needed revision" or "confirmed a successful clip placement." The explicit adjudication method here is not detailed; it appears to be based on the surgeon's intraoperative assessment aided by the FL800, potentially re-evaluated with standard postoperative imaging (though not explicitly stated for aneurysm cases as it was for bypasses).
   • This is not a traditional consensus-based adjudication method but rather a comparison to standard clinical follow-up diagnostics.

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, a multi-reader multi-case (MRMC) comparative effectiveness study was not done. This study focused on the standalone performance of the device assist for a single surgeon, with surgical decisions confirmed by independent follow-up diagnostics. The document does not provide data on improvement with AI vs. without AI assistance for human readers.

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

   • The study was designed to evaluate the device with a human-in-the-loop (the surgeon) during actual surgical procedures.
   • The performance described is the qualitative assessment made by the surgeon, enabled by the device, whose findings were later confirmed by other diagnostic methods. There is no isolated "algorithm only" performance reported.

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

   • Clinical Outcomes/Diagnostic Confirmation: The primary ground truth for bypass cases was confirmation by sonogram and postoperative angiography. For aneurysm cases, it was the surgical outcome and need for revision based on intraoperative assessment, later confirmed by successful clip placement.
   • This can be categorized as a combination of diagnostic imaging confirmation and clinical outcome.

7. The sample size for the training set:

   • The document does not mention a training set in the context of an algorithm. The Leica FL800 is a hardware accessory for a surgical microscope that visualizes ICG fluorescence. It doesn't appear to incorporate an "algorithm" in the modern sense that requires a training set for machine learning.
   • The "validation of concept" in Dresden with unspecified cases might be considered a preliminary testing phase leading to hardware refinements, but not a training set for an AI model.

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

   • Not applicable, as there is no mention of a "training set" for an algorithm.

Ask a specific question about this device