(113 days)
The noninvasive INVOS® 3100A Cerebral Oximeter should be used in adults as an adjunct monitor of trends in regional hemoglobin oxygen saturation of blood in the brain of an individual. Because INVOS values are relative within an individual, the INVOS should not be used as the sole basis for decisions as to diagnosis or therapy. The value of data from the INVOS has not been demonstrated in disease states.
The principles of operation of the cerebral oximeter system are based on the assumption that hemoglobin exists in two principal forms in the blood: oxygenated hemoglobin (HbO2) and reduced hemoglobin (Hb). Functional oxygen saturation (SO2) is defined as the ratio of oxyhemoglobin (HbO2) to total hemoglobin (HbO2 + Hb) and is commonly presented as a percentage. Since oxygenated and reduced hemoglobin are different colors and absorb light as a known function of wavelength, selected wavelengths of light can be used to assess the relative percentage of these two constituents. This fundamental approach of assessing the color of blood using various wavelengths of light to measure hemoglobin oxygen saturation trends is used in all currently marketed oximetry systems. A disposable sensor of medical grade materials is applied to the patient's forehead. The sensor incorporates a light source and two return signal detectors at different predetermined distances from the light source. The signal detector nearest the light source (3 cm) is considered the "shallow detector" and the further detector from the light source (4 cm) the "deep detector." While the light reaching the deep detector has sampled about the same amount of skin, scalp, and skull as the light reaching the shallow detector, it has sampled more brain. This difference is used to help separate out the brain signal and suppress tissue anatomical differences in patients. The additional information unique to the deep signal return is predominately from brain tissue blood which is composed mostly of venous blood. The information contained in the shallow and deep signal returns is processed by an algorithm to measure changes in hemoglobin oxygen saturation in a small region of tissue beneath the sensor, predominately in the brain. The SomaSensor is connected to a preamplifier which is placed close to the patient and amplifies the rSO2 signal. The signal is then carried to a display unit where the values and trends are displayed on the screen. The display unit controls all functions of the system with selections made by keys with onscreen labels. The system will operate for up to 20 minutes on battery, enabling patient transport without loss of data.
The INVOS 3100A Cerebral Oximeter is designed to monitor trends in regional hemoglobin oxygen saturation in the brain. The device's acceptance criteria and the studies proving its performance are detailed below. It's important to note that this device is intended as an adjunct monitor and not for sole diagnostic or therapeutic decisions.
1. Table of Acceptance Criteria and Reported Device Performance
| Acceptance Criteria Category | Specific Metric | Acceptance Criteria | Reported Device Performance |
|---|---|---|---|
| Trend Agreement (Hypoxia Study) | Mean individual r^2 (correlation between fSO2 and rSO2) | A high correlation value | 0.947 (range 0.805 to 0.991) |
| Trend Accuracy (Hypoxia Study) | Ability to accurately measure trends (combined bias and standard deviation) | Within ±X% (implied to be clinically acceptable, often ±5%) | Within ±4.9% |
| Trend Measurement Correlation Coefficient (Hypoxia Study) | r (between fSO2 and rSO2) | A high correlation value | 0.935 |
| Bias (Hypoxia Study) | Overall mean bias between fSO2 and rSO2 index | Low bias (e.g., close to 0) | 1.33 |
| Standard Deviation of Absolute Difference (Hypoxia Study) | Between fSO2 and rSO2 index for individuals | Low standard deviation | Averaged 3.08 |
| Transition Accuracy (Hypoxia Study - Systemic Hypoxia) | During changes in SaO2 (up to 27%) with constant CO2 | Within X% (implied clinically acceptable, e.g., ±5.5%) | Within 5.5% |
| Transition Accuracy (Hypoxia Study - Cerebral Blood Flow Changes) | During changes in CO2 (4-10 mmHg) with constant SaO2 | Within X% (implied clinically acceptable, e.g., ±5.5%) | Within 5.5% |
| Correlation with MCAVm (CEA Study) | Correlation between changes in rSO2 index and MCAVm during cross-clamp | A strong correlation value | 0.806 |
| Detection of Ischemia (CEA Study) | INVOS detection of oxygenation changes preceding EEG changes | Statistically significant precedence | p 95%) |
| Patient Safety (Both Studies) | Absence of adverse reactions or skin irritation | No instances observed | No instances of irritation or adverse reactions observed |
2. Sample Sizes Used for the Test Set and Data Provenance
- Hypoxia Study: 30 volunteers.
- Data Provenance: Not explicitly stated, but implies a prospective study conducted in a controlled lab setting where hypoxia and hypercapnia were induced. The demographics (skin tone, age, gender) are provided, but country of origin is not specified.
- Carotid Endarterectomy (CEA) Study: 27 patients (one operated twice for both left and right CEAs, and one additional subject for operational reliability data).
- Data Provenance: Not explicitly stated, but implies a prospective study conducted during surgical procedures. All patients were Caucasian. Country of origin is not specified.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications
- Hypoxia Study: The "ground truth" (fSO2) was calculated from arterial and jugular venous blood oxygen saturations, measured off-line on a cooximeter. While a cooximeter is a device, its results are typically interpreted and validated by trained personnel. No specific "experts" are mentioned for establishing ground truth from this data itself, beyond the standard medical laboratory practices for cooximetry.
- CEA Study:
- EEG Changes: "evaluated by a trained observer using 10-channel analog recordings." The specific qualifications of this "trained observer" are not provided (e.g., neurophysiologist, technician, years of experience).
- MCAVm: Measured by transcranial Doppler (TCD), which is a device-based measurement, implying a trained technician/physician to operate and interpret.
- The "gold standard" for cerebral ischemia in this context is implicitly a combination of TCD and EEG changes.
4. Adjudication Method for the Test Set
- The document does not describe a formal adjudication method (like 2+1 or 3+1 consensus) for the ground truth in either study. Ground truth appears to be based on direct physiological measurements (blood samples, TCD, EEG) interpreted by standard clinical practice or single trained observers (for EEG).
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
- No MRMC comparative effectiveness study was done. The focus was on comparing the device's output to established physiological measurements and other monitoring devices (cooximeter, TCD, EEG), not on how human readers' performance improved with AI assistance.
6. Standalone Performance (Algorithm Only Without Human-in-the-Loop Performance)
- Yes, the performance metrics (trend agreement, accuracy, correlation, bias, operational time) are all measurements of the device's standalone performance. The INVOS 3100A Cerebral Oximeter operates autonomously to provide rSO2 index values. The studies evaluate how well these autonomously generated values correlate with or detect physiological changes measured by other means.
7. Type of Ground Truth Used
- Hypoxia Study:
- Physiological measurements: Functional oxygen saturation (fSO2) derived from arterial and jugular venous blood samples, measured off-line on a cooximeter. This is a direct physiological measurement.
- CEA Study:
- Physiological measurements/clinical indicators: Changes in mean middle cerebral artery flow velocity (MCAVm) as measured by transcranial Doppler (TCD) and EEG changes evaluated by a trained observer. These serve as indicators of cerebral function and ischemia.
8. Sample Size for the Training Set
- The document does not explicitly mention a "training set" in the context of machine learning or AI models. The INVOS 3100A Cerebral Oximeter operates based on spectrophotometric principles and an algorithm to process signals from its sensors. The development of this algorithm itself would have involved internal validation and calibration, but details on a separate, dedicated "training set" in the modern sense of AI algorithm development are not provided. The non-clinical testing described likely encompasses internal verification and validation of the device's fundamental function and algorithms.
9. How the Ground Truth for the Training Set Was Established
- As a "training set" is not explicitly mentioned, the method for establishing its ground truth is also not detailed. The device's underlying principles (spectrophotometry) are well-established. The algorithm's development would have relied on biophysical models and experimental data to accurately translate light absorption into regional oxygen saturation, consistent with established physiological knowledge.
§ 870.2700 Oximeter.
(a)
Identification. An oximeter is a device used to transmit radiation at a known wavelength(s) through blood and to measure the blood oxygen saturation based on the amount of reflected or scattered radiation. It may be used alone or in conjunction with a fiberoptic oximeter catheter.(b)
Classification. Class II (performance standards).