Evaluation Of Detector Position and Light Fluence Distribution Using an Infrared Navigation System during Pleural Photodynamic Therapy

Abstract

Photodynamic therapy (PDT) has been used to treat malignant pleural mesothelioma. Current practice involves delivering light to a prescribed light fluence with a point source, monitored by eight isotropic detectors inside the pleural cavity. An infrared (IR) navigation system was used to track the location of the point source throughout the treatment. The recorded data was used to reconstruct the pleural cavity and calculate the light fluence to the whole cavity. An automatic algorithm was developed recently to calculate the detector positions based on recorded data within an hour. This algorithm was applied to patient case studies and the calculated results were compared to the measured positions, with an average difference of 2.5 cm. Calculated light fluence at calculated positions were compared to measured values. The differences between the calculated and measured light fluences were within 14% for all cases, with a fixed scattering constant and a dual correction method. Fluence surface histogram (FSH) was calculated for photofrin-mediated PDT to be able to cover 80% of pleural surface area to 50 J/cm²(83.3% of 60 J/cm²). The study demonstrates that it will be possible to eliminate the manual measurement of the detector positions, reducing the patient’s time under anesthesia.

INTRODUCTION

As a rare cancer of pleura, malignant pleural mesothelioma (MPM) is aggressive and lethal, which results in a typical median survival of 9–17 months irrespective of stage (1, 2). Although, in theory, complete surgical resection can be an effective treatment method, it is hardly achieved due to the common diffusely spread of MPM throughout the hemithorax (3). Thus, multimodality approaches involving surgery are commonly employed for treating MPM (4). Photodynamic therapy (PDT), one of the progressively more important intraoperative modalities, has been demonstrated to be able to improve local tumor control and overall survival without a cumulative toxicity (5–7).

PDT delivers non-ionizing irradiation of a specific wavelength that activates photosensitizer drug injected into a patient before surgery to produce singlet oxygen that can trigger various anti-cancer activities (8–10). Different factors, including drug concentration, light fluence, and oxygen availability, can essentially affect the efficacy of PDT. To optimize PDT, it is crucial to ensure the proper delivery of light by measuring the actual fluence in the treatment target area while keeping operation time as short as possible (11, 12). In the existing clinical protocol, eight detectors were placed at discrete locations inside the pleural cavity to measure the light fluence rate (mW/cm²) as well as monitor the accumulated light fluence (J/cm²) throughout the treatment. An infrared navigation system was used to track the light source position during PDT to calculate the light fluence distribution for the entire cavity.

In this study, the light fluence distribution calculation was explored for a set of clinical cases involving pleural PDT between 2014 and 2018. Different techniques for the determination of detector positions and different light calculation methods were investigated and assessed. This study aims to develop and validate an algorithm to extrapolate the detector positions after PDT and eliminate the existing position measurement procedure before treatment. This can effectively reduce the overall anaesthetic time.

MATERIALS AND METHODS

Patient diagnosis and pleural PDT

A clinical trial (ClinicalTrials.gov Identifier: NCT02153229) at the Hospital of the University of Pennsylvania was approved for MPM patients to be treated through surgical resection with interoperative PDT. The PDT technique and protocol were described in depth in previous studies (13, 14). The photosensitizer drug, Photofrin (Porfimer Sodium), was administered intravenously at 2 mg/kg body weight, followed 24 h later by illumination using laser light of 630 nm wavelength with typically laser power of 5W (power setting of 6W on laser) (13). The prescribed light fluence was 60 J/cm². The light was delivered through a bare fiber enclosed in an endotracheal tube connected to a diode laser system (Modulight, Tampere, Finland). Two versions of the treatment delivery wand were utilized in the cases involved, and a detailed illustration can be found elsewhere (14, 15). The wand and the pleural cavity were filled up with diluted Intralipid solution for light scattering. Eight isotropic detectors (Medlight SA, Ecublens, Switzerland) were sutured inside the cavity prior to PDT to measure the light fluence rate throughout the treatment, and the measured light fluence (ϕ_measured) for the entire treatment were obtained afterward. The detectors were located discretely: apex, posterior chest wall (PCW), anterior chest wall (ACW), posterior sulcus (PS), anterior sulcus (AS), posterior mediastinum (PM), pericardium (Peri), diaphragm (Diaph).

Infrared (IR) navigation system and pleural cavity geometry reconstruction

A commercial IR navigation system (Polaris, NDI, Waterloo, Canada), consisting of cameras and passive reflective markers, was used for light source tracking during pleural PDT (14–16). The camera system includes a pair of cameras to form a stereo-cameras system that can track the passive reflective markers mounted on the treatment delivery wand by measuring the reflected light at a rate of 20–60 Hz. Tracking reflective markers allows monitor of the tip position of the wand, i.e., the laser source position, continuously throughout the PDT. This position is given by the 3D Cartesian coordinates (x, y, and z) and the orientation (Q0, Q1, Q2, Q3). The position data can be visualized in real-time and recorded for cavity contour reconstruction and post-treatment data processing. The accuracy of the system is around 0.5 mm in 3D, with a maximum active detection volume of approximately 205 × 186 × 147 cm³, which is the optimal range for use during the pleural cavity treatment determined from previous clinical cases. The IR camera system is positioned on the ceiling in the operation room before treatment. For the existing cases, a different rod with reflective markers was used to measure the position of the eight isotropic detectors (p_measured) within the pleural cavity prior to irradiation.

The pleural cavity geometry was reconstructed using the recorded data obtained within the cavity and served as the target surface for light fluence calculation. The reconstruction algorithm was demonstrated thoroughly in a previous study (14). The Cartesian coordinates utilized in raw data were transformed into spherical coordinates, and the boundary was obtained using the outermost data points. The pleural cavity geometry reconstructed using treatment data is presented in Fig. 1 with the eight isotropic detectors shown.

Figure 1.

Calculated detector positions inside a reconstructed patient cavity contour, both obtained from recorded laser tip position data.

Calculation of isotropic detector positions

For all existing clinical cases, the position of the eight isotropic detectors, sutured inside the pleural cavity, was measured manually by the physician prior to treatment. This procedure was performed by pointing a separate wand tip to the discrete detectors one at a time and recording each position coordinates separately. This is a feasible and relatively accurate method for most cases. However, some drawbacks are identified from previous treatment (15). The physician may find it hard to locate the detector with the navigation system due to the cavity structure or blocked view from the camera to the reflective markers. Another issue is that, in some scenarios, patients were repositioned during PDT, and the shift in detectors rendered the measured values invalid for part of the treatment. More importantly, this step undesirably extended operation time and complicated the overall treatment procedure. Thus, for a more efficient treatment procedure, an automatic algorithm for post-treatment detector positions computation has been developed, which allows the manual measuring step to be eliminated.

The schematic of the algorithm is illustrated in Fig. 2. The algorithm is based on the idea that the eight isotropic detectors only get illuminated when the light source gets close enough to them, and the corresponding light fluence rate is recorded. The recorded data is then used to process the measured light fluence (cumulative fluence data), as shown by the solid red lines in Fig. 3–6. The regions where sharp increases occur indicate that the detector is illuminated for the period. These features are highlighted in grey in Fig. 3(h). The detectors are relatively distant from each other, which means that typically only one detector is illuminated at each time t. Occasionally, more than one detector is illuminated simultaneously, but the specific one which detects a significantly higher fluence rate value can be identified clearly. Once an illuminated detector is identified at time t, the algorithm will find the corresponding light source position. The light fluence rate at this detector and the light source position are recorded. If no valid position data is recorded for time t, the algorithm will skip to time t+1. All the stored data are categorized according to different detectors to construct sets of position data together with the fluence rate for each detector. The detector positions can then be extrapolated by processing the data sets. The calculated detector positions (p_calculated) were compared to the measured positions with the shifts calculated. All 11 clinical cases were analyzed using both calculated and measured positions for further validation.

Figure 2.

Schematics of the automatic algorithm to calculate isotropic detector positions using treatment data. (a) Identify laser tip positions when detector gets illuminated and group according to different detectors. (b) For each detector, calculate the extrapolated position based on weighted average value.

Figure 3.

Measured (red solid lines) light fluence data (ϕ_measured) obtained by the isotropic detectors over the course of treatment along with calculated (blue dashed lines) light fluence (ϕ_calculated) using the primary component (Eq. 1) for patient #37 based on detector positions (p_calculated) computed by the algorithm plotted for eight detector locations: (a) posterior sulcus (PS) (b) posterior mediastinum (PM) (c) posterior chest wall (PCW) (d) apex (e) pericardium (Peri) (f) anterior chest wall (ACW) (g) anterior sulcus (AS) (h) diaphragm (Diaph).

Figure 6.

Measured (red solid line) light fluence data (ϕ_measured) obtained by the isotropic detectors over the course of treatment along with calculated (blue dashed line) light fluence (ϕ_calculated) using the primary component with fixed scattered light and the dual correction method (Eq. 2) for patient #37 based on calculated detector positions (p_calculated) computed by the algorithm plotted for eight detector locations: (a) posterior sulcus (PS) (b) posterior mediastinum (PM) (c) posterior chest wall (PCW) (d) apex (e) pericardium (Peri) (f) anterior chest wall (ACW) (g) anterior sulcus (AS) (h) diaphragm (Diaph).

Light fluence calculations and statistical analysis

With the reconstructed pleural cavity geometry and position data of the light source throughout the treatment obtained from the IR navigation system, the distance from the light source to the pleural cavity surface, as well as the light fluence rate to each point on the cavity, can be calculated. The light fluence rate is the summation of a primary and a scattered component, i.e., direct and scattered light (14). The primary component, ϕ_primary, can be calculated by

$${\varphi }_{\text{primary}}=\frac{S}{4\pi r^2}$$ (1)

where S (mW) is the light source power and r (cm) is the distance from the light source to the point of interest. To account for the scattered component, a constant b (mWcm⁻²) is considered. It served as the scattered component added in for total light fluence calculation, i.e., ϕ_scatter = b. A dual correction factor (CF(t)) was applied to the entire calculated light fluence rate to further improve the agreement between measured and calculated results. The value of CF at a given time is evaluated using the measured and calculated light fluence of the detector with the largest sum fluence at that time (14). With the scattered component and CF considered, the total light (ϕ_calculated) is calculated using the following equation

$$\varphi \left(r,t\right)=\left(\frac{S}{4\pi r{\left(t\right)}^2}+b\right)\cdot CF\left(t\right)$$ (2)

To compare and evaluate different light fluence calculation methods, the percentage difference between the measured and calculated light fluence was calculated at the end of treatment at each detector position.

RESULTS

Table 1 summarizes percentage of good data (i.e., light source position data when more than three of the reflective markers on the wand were successfully tracked by the IR camera during PDT treatment), values of scattering component (b), surface area and volume of pleural cavity obtained from the navigation data for each case. Table 2 shows extrapolated detector positions compared to the measured detector positions with the shifts calculated. Table 3 shows the percent error between measured and calculated light fluence at the end of treatment tracked by the navigation system using (a) primary component only, (b) primary and scattering component, (c) primary component with dual CF, and (d) primary and scattering component with dual CF. Table 4 summarizes the uniformity of the light fluence distribution in patients as quantified as the standard deviation of the mean and the variation of the standard deviation.

Table 1.

Summary of percentage of good data, pleural cavity surface area, volume, and scattering component (b) for each patient.

Case No.	Percentage of good data (%)	Area	Volume	Power (mW)	b (mW/cm2)
04	36.9%	1496	7790	5200	7.2
08	38.3%	1520	7010	5200	7.0
12	19.5%	886	2742	5040	7.5
14	37.2%	1710	8192	5200	7.5
16	45.0%	1158	6095	5200	7.5
17	18.7%	1447	7618	5200	6.5
18	25.0%	1766	8103	5512	7.5
20	22.3%	1262	6308	5200	7.0
27	36.3%	1771	8113	5200	7.2
37	27.9%	1289	6069	5400	7.8
38	24.7%	1151	5699	5400	7.8
Average	30.2±8.9%	1405±284	6704±1609	5250±132	7.3±0.4

Table 2.

Shifts between measured and extrapolated detector locations in each coordinate direction (∆x, ∆y, and ∆z) and total distance (d) for each detector site (Diaphram (Diaph), Posterior sulcus (PS), Anterior sulcus (AS), Posterior chest wall (PCW), Anterior chest wall (ACW), Pericardium (Peri), and Apex) per patient and averaged across all patient cases for each site. Shifts are in units of cm. (For 004, PS and AS, there is no navigation data to extrapolate these detector positions because the NDI camera was blocked at the delivery time.)

Detector	Δx(cm)	Δy(cm)	Δz(cm)	d(cm)	Δx(cm)	Δy(cm)	Δz(cm)	d(cm)
	Case 004				Case 008
Diaph	0.71	−0.20	2.42	2.53	−1.84	0.88	−0.10	2.04
PS	-	-	-	-	−0.32	−0.35	−1.06	1.16
AS	-	-	-	-	−0.04	0.37	−0.94	1.01
PM	0.21	0.84	0.22	0.89	1.17	−0.44	1.66	2.08
PCW	−0.01	0.02	−0.03	0.04	−0.90	2.12	0.31	2.33
ACW	1.53	−0.13	−0.18	1.55	0.96	−0.16	1.64	1.91
Peri	0.89	0.19	1.78	2.00	−2.24	0.71	2.34	3.31
Apex	−0.02	−0.54	3.73	3.77	0.80	2.09	0.93	2.42
Average	0.55±0.61	0.030±0.47	1.32±1.58	1.80±1.30	0.30±1.28	0.65±1.02	0.60±1.26	2.03±0.73
	Case 012				Case 014
Diaph	1.18	1.75	2.14	3.00	0.40	−2.22	0.62	2.34
PS	1.06	−1.19	1.14	1.96	−1.22	1.49	−0.05	1.93
AS	−1.43	−0.18	3.39	3.68	0.14	2.74	1.35	3.06
PM	1.78	2.99	−1.56	3.81	1.39	1.53	−0.29	2.09
PCW	−3.31	−1.97	−0.95	3.97	−2.72	−2.19	−1.24	3.70
ACW	0.67	0.76	−1.22	1.59	−0.12	−0.24	−1.18	1.21
Peri	0.54	−1.44	1.66	2.27	−0.03	2.56	−1.74	3.09
Apex	−1.67	−2.89	−2.15	3.97	3.62	−0.64	−2.15	4.26
Average	0.15±1.77	−0.27±1.99	0.31±2.03	3.03±0.97	0.18±1.85	0.38±1.99	0.59±1.20	2.71±1.00
	Case 016				Case 017
Diaph	0.40	−2.22	0.62	2.34	−1.91	−0.29	2.52	3.17
PS	−1.22	1.49	−2.36	3.05	2.11	−2.98	0.63	3.71
AS	0.14	2.74	1.35	3.06	1.17	0.20	0.09	1.19
PM	−0.67	0.87	−1.78	2.09	1.29	1.76	−4.69	5.17
PCW	2.08	−0.89	−0.20	2.27	−2.32	2.71	0.10	3.57
ACW	−0.12	−0.24	−1.18	1.21	0.45	6.08	−0.35	6.11
Peri	−0.42	0.02	−1.74	1.79	0.50	1.61	1.62	2.34
Apex	3.62	−0.64	−2.15	4.26	−0.53	0.57	0.04	0.78
Average	0.48±1.60	0.14±1.54	0.93±1.37	2.51±0.94	0.09±1.57	1.21±2.61	0.01±2.12	3.25±1.83
	Case 018				Case 020
Diaph	−0.07	−0.50	−1.46	1.55	0.29	−2.31	2.31	3.28
PS	1.68	−0.90	1.00	2.15	−1.22	1.49	−0.05	1.93
AS	1.16	−1.31	1.11	2.07	−1.25	2.10	1.00	2.64
PM	−1.00	−0.65	−2.52	2.79	−0.47	−0.13	0.27	0.55
PCW	2.53	−0.87	2.31	3.54	0.03	−0.53	−1.22	1.34
ACW	−0.05	0.81	−0.39	0.90	−1.08	2.49	0.59	2.78
Peri	−0.81	1.42	1.69	2.35	−0.42	0.02	−4.55	4.57
Apex	0.60	−1.26	1.62	2.14	3.62	−0.64	−2.15	4.26
Average	0.51±1.23	−0.41±0.99	0.42±1.70	2.19±0.78	0.06±1.59	0.31±1.61	0.48±2.13	2.67±1.38
	Case 027				Case 037
Diaph	0.96	0.03	1.62	1.89	0.24	0.04	−1.91	1.93
PS	0.98	−0.33	−0.35	1.09	0.47	−0.32	1.55	1.65
AS	−0.71	0.60	1.75	1.99	2.90	1.87	−2.11	4.04
PM	1.57	−1.17	−1.89	2.72	−2.15	−0.55	−1.24	2.54
PCW	−1.95	−0.59	−3.68	4.21	−0.41	1.73	0.80	1.95
ACW	0.13	−0.41	1.37	1.44	−1.97	0.98	0.01	2.21
Peri	1.33	−0.95	5.16	5.41	1.86	1.20	1.63	2.74
Apex	−1.12	−0.85	−0.46	1.48	−0.03	0.69	−1.42	1.58
Average	0.15±1.28	−0.46±0.57	0.44±2.68	2.53±1.52	0.11±1.72	0.70±0.91	0.34±1.53	2.33±0.80
	Case 038				Averaged across all cases
Diaph	0.43	1.18	0.52	1.36	1.20±2.10	0.47±1.34	0.80±1.41	2.84±1.33
PS	−2.54	−2.53	0.00	3.58	0.01±1.44	0.30±1.56	0.06±1.09	2.10±0.97
AS	−0.25	−0.29	−1.02	1.09	0.73±1.95	0.32±2.05	0.55±1.62	3.09±1.02
PM	−0.57	0.77	0.46	1.07	0.06±0.91	0.72±2.14	0.05±1.13	2.18±1.43
PCW	−0.27	3.65	−1.72	4.04	0.17±1.23	0.81±1.35	0.54±1.54	2.17±1.25
ACW	0.94	−2.36	−0.50	2.59	0.35±1.35	0.44±1.26	1.07±1.72	2.40±1.27
Peri	1.98	1.32	−0.16	2.39	0.21±1.23	0.54±1.22	0.97±2.91	3.18±1.32
Apex	3.04	−1.17	−0.09	3.26	0.07±1.02	0.35±1.39	0.84±1.54	2.31±0.64
Average	0.35±1.70	0.07±2.08	0.31±0.75	2.42±1.16

Table 3.

Percent error between measured and calculated light fluence at the end of treatment tracked by navigation system using (a) primary component only, (b) primary and scattering component, (c) primary component with CF, and (d) primary and scattering component with CF. (For 004, PS and AS, there is no navigation data to make comparisons because the NDI camera was blocked at the delivery time.)

(a) Primary component only
Case no.	Diaph	PS	AS	PM	PCW	ACW	Peri	Apex	Avg.
004	52.2%	-	-	48.4%	56.5%	56.4%	50.3%	37.1%	50.2%±7.2%
008	40.7%	50.4%	32.8%	42.0%	51.4%	39.8%	47.3%	40.8%	43.2%±6.2%
012	59.5%	42.1%	52.3%	55.2%	45.8%	34.9%	41.0%	33.8%	45.6%±9.4%
014	55.0%	54.3%	42.4%	56.7%	61.3%	53.1%	32.7%	60.7%	52.0%±9.7%
016	56.5%	45.2%	43.7%	47.9%	53.4%	60.3%	59.9%	27.9%	49.4%±10.8%
017	35.7%	29.7%	56.9%	49.2%	54.9%	35.9%	43.1%	43.8%	43.7%±9.6%
018	29.4%	53.0%	54.3%	49.3%	59.5%	61.1%	49.3%	56.6%	51.6%±9.9%
020	59.3%	55.0%	51.0%	49.3%	56.7%	56.0%	61.7%	59.4%	56.1%±4.2%
027	33.4%	43.3%	55.4%	32.8%	58.9%	47.5%	46.1%	59.5%	47.1%±10.5%
037	39.1%	49.8%	48.8%	51.0%	44.8%	53.0%	44.7%	46.4%	47.2%±4.4%
038	49.7%	53.9%	52.0%	58.4%	61.9%	52.9%	56.8%	51.4%	54.6%±4.1%
Avg.	46.4% ±11.0%	47.7% ±7.9%	49.0% ±7.4%	49.1% ±7.1%	55.0% ±5.8%	50.1% ±9.3%	48.5% ±8.6%	47.0% ±11.4%
(b) Primary and scattering component
004	9.7%	-	-	8.2%	10.7%	11.9%	4.7%	10.8%	9.3%±2.6%
008	9.8%	10.6%	8.4%	0.4%	9.2%	6.2%	2.3%	7.9%	6.9%±3.7%
012	8.4%	6.4%	3.6%	10.7%	4.5%	8.3%	4.1%	2.0%	6.0%±3.0%
014	4.9%	2.2%	1.3%	6.9%	10.9%	3.5%	5.5%	6.4%	5.2%±3.0%
016	7.2%	6.6%	5.0%	7.6%	3.9%	11.1%	7.9%	11.2%	7.6%±2.6%
017	13.6%	9.8%	10.9%	6.0%	6.6%	5.8%	4.4%	12.6%	8.7%±3.5%
018	4.9%	7.7%	7.8%	8.6%	7.9%	5.6%	5.3%	13.5%	7.7%±2.7%
020	9.1%	12.9%	10.9%	8.9%	6.9%	5.7%	2.2%	9.2%	8.2%±3.3%
027	6.7%	7.0%	4.9%	1.8%	3.5%	1.7%	4.3%	12.8%	5.3%±3.6%
037	10.0%	6.1%	2.4%	12.8%	1.1%	6.9%	7.4%	2.7%	6.2%±4.0%
038	6.3%	2.1%	11.2%	12.7%	3.0%	5.2%	6.4%	0.7%	6.0%±4.2%
Avg.	8.2% ±2.6%	7.1% ±3.4%	6.6% ±3.7%	7.7% ±3.9%	6.2% ±3.3%	6.5% ±3.0%	5.0% ±1.8%	8.2% ±4.6%
(c) Primary and scattering component with fixed power and scattering component
004	10.4%	-	-	10.2%	11.1%	9.3%	5.8%	9.5%	9.4%±1.9%
008	11.8%	13.5%	10.9%	3.7%	5.4%	9.9%	0.6%	10.3%	8.3%±4.5%
012	6.9%	8.7%	6.8%	14.8%	7.0%	4.8%	7.5%	1.3%	7.2%±3.8%
014	8.6%	2.0%	3.4%	10.3%	6.7%	7.7%	1.2%	11.3%	6.4%±3.8%
016	2.6%	1.4%	9.4%	12.9%	7.7%	7.6%	3.1%	16.4%	7.6%±5.2%
017	3.2%	18.1%	2.8%	15.0%	16.1%	5.1%	6.1%	20.9%	10.9%±7.3%
018	8.7%	13.0%	2.3%	12.3%	12.1%	10.3%	9.2%	6.8%	9.3%±3.5%
020	12.3%	15.8%	7.1%	10.7%	4.3%	8.3%	1.3%	11.8%	9.0%±4.7%
027	7.7%	5.2%	6.5%	3.0%	5.3%	2.1%	5.6%	11.1%	5.8%±2.8%
037	3.7%	14.7%	5.2%	19.6%	7.5%	13.5%	1.0%	9.5%	9.3%±6.2%
038	14.6%	6.3%	19.9%	5.3%	6.9%	11.5%	1.7%	8.6%	9.4%±5.8%
Avg.	8.2% ±3.9%	9.9% ±6.0%	7.4% ±5.2%	10.7% ±5.1%	8.2% ±3.5%	8.2% ±3.3%	3.9% ±3.0%	10.7% ±5.0%
(d) Primary component with dual correction (CF)
004	3.9%	-	-	3.6%	21.4%	13.5%	11.9%	30.1%	14.1%±10.3%
008	19.2%	9.2%	15.8%	16.7%	3.4%	18.3%	1.3%	15.8%	12.5%±6.9%
012	6.11%	19.7%	17.1%	13.5%	2.1%	13.2%	24.5%	18.5%	14.3%±7.3%
014	13.3%	8.9%	33.8%	4.3%	9.8%	7.9%	19.2%	5.0%	12.8%±9.7%
016	21.9%	16.2%	13.8%	2.7%	13.3%	2.1%	1.6%	11.9%	10.4%±7.5%
017	16.0%	3.5%	18.5%	17.5%	16.6%	6.6%	16.8%	15.5%	13.9%±5.6%
018	14.6%	1.3%	2.2%	12.6%	11.5%	16.9%	10.8%	16.8%	10.8%±6.0%
020	14.9%	20.0%	26.4%	6.6%	15.7%	10.2%	1.8%	10.8%	13.3%±7.7%
027	5.6%	17.5%	3.1%	14.5%	8.3%	7.0%	22.9%	22.8%	12.7%±7.8%
037	4.3%	8.2%	13.9%	3.7%	15.0%	5.9%	6.6%	14.0%	9.0%±4.6%
038	8.3%	6.6%	18.6%	17.9%	26.5%	17.4%	16.1%	20.8%	16.5%±6.4%
Avg.	11.6% ±6.3%	11.1% ±6.8%	16.3% ±9.5%	10.3% ±6.2%	13.1% ±7.2%	10.8% ±5.4%	12.1% ±8.5%	16.5% ±6.6%
(e) Primary and scattering component with dual correction (CF)
004	3.1%	-	-	2.1%	9.5%	7.1%	7.5%	10.8%	6.7%±3.5%
008	5.3%	1.5%	6.5%	8.4%	1.4%	4.3%	6.4%	3.7%	4.7%±2.5%
012	4.5%	3.3%	1.5%	6.3%	1.3%	9.8%	10.6%	9.6%	5.9%±3.8%
014	9.2%	6.8%	8.1%	3.4%	5.0%	6.4%	3.3%	2.7%	5.6%±2.4%
016	6.8%	8.2%	9.4%	4.4%	1.9%	6.5%	1.6%	3.8%	5.3%±2.9%
017	7.8%	3.3%	4.4%	11.9%	9.1%	4.8%	2.6%	5.7%	6.2%±3.2%
018	5.6%	2.7%	3.1%	6.8%	7.5%	7.3%	9.6%	1.0%	5.5%±2.9%
020	7.1%	3.0%	10.8%	8.7%	3.4%	0.9%	0.5%	3.2%	4.7%±3.7%
027	3.6%	8.6%	7.9%	7.0%	8.8%	4.0%	8.3%	2.4%	6.3%±2.6%
037	2.8%	1.4%	4.3%	7.9%	3.2%	3.5%	3.4%	1.1%	3.5%±2.1%
038	4.6%	2.1%	8.0%	7.9%	10.1%	8.3%	8.0%	5.7%	6.8%±2.5%
Avg.	11.6% ±6.3%	11.1% ±6.8%	16.3% ±9.5%	10.3% ±6.2%	13.1% ±7.2%	10.8% ±5.4%	12.1% ±8.5%	16.5% ±6.6%

Table 4.

Summary of uniformity of fluence distribution for photofrin-mediated PDT cases.

Case no.	Standard deviation (% difference)	Variation of std. deviation (% difference)
004	10.2	14.3
008	9.2	15.5
012	5.6	13.0
014	11.0	22.3
016	5.0	14.5
017	14.2	26.2
018	12.6	21.4
020	8.5	15.6
027	12.7	28.1
037	9.1	22.9
038	4.6	16.2
Average	9.3%±3.2%	19.1%±5.3%

The results for a representative case (patient #37), chosen based on key features (well defined primary fluence rates increase) and recent treatment date, are presented. Figure 3 compares the calculated and measured light fluence measured by the eight isotropic detectors using the primary fluence (Eq. 1) at these (calculated) detector positions. Figures 4 and 5 compare the calculated (ϕ_calculated) and measured light fluences (ϕ_measured) using the primary and scatter fluence (Eq. 2, CF = 1) at these calculated (p_calculated) and measured detector positions (p_measured), respectively. Figure 6 compares the calculated and measured light fluences using the primary and scatter fluence with the dual correction factor (Eq. 2) at calculated detector positions. Figure 7 shows the light fluence distribution at the end of PDT treatment of 8 patients for whom we can track the light delivery for the entirety of PDT treatment (until 60 J/cm²) using the primary and scatter dose plus dual correction. The symbol (“x”) marked the location of the isotropic detectors. Figure 8 shows the fluence-surface histogram for the eight patients evaluated in Fig. 7.

Figure 4.

Measured (red solid lines) light fluence data (ϕ_measured) obtained by the isotropic detectors over the course of treatment along with calculated (blue dashed lines) light fluence (ϕ_calculated) using the primary component with fixed scattered light (Eq. 2, CF = 1) for patient #37 based on calculated detector positions (p_calculated) computed by the algorithm plotted for eight detector locations: (a) posterior sulcus (PS) (b) posterior mediastinum (PM) (c) posterior chest wall (PCW) (d) apex (e) pericardium (Peri) (f) anterior chest wall (ACW) (g) anterior sulcus (AS) (h) diaphragm (Diaph).

Figure 5.

Measured (red solid lines) light fluence data (ϕ_measured) obtained by the isotropic detectors over the course of treatment along with calculated (blue dashed lines) light fluence (ϕ_calculated) using the primary component with fixed scattered light (Eq. 2, CF = 1) for patient #37 based on measured detector positions (p_measured) computed by the algorithm plotted for eight detector locations: (a) posterior sulcus (PS) (b) posterior mediastinum (PM) (c) posterior chest wall (PCW) (d) apex (e) pericardium (Peri) (f) anterior chest wall (ACW) (g) anterior sulcus (AS) (h) diaphragm (Diaph), for comparison with the result based on calculated detector position (Fig.4).

Figure 7.

Fluence distribution maps for 8 patients. The 3D geometry is unwrapped and displayed on a 2D surface plot of z-depth (in cm) and unwrapped angle (in degree). The locations of the calculated isotropic detectors are indicated by ‘x’ symbols.

Figure 8.

Cumulative light fluence surface histogram (DSH) for eight patients. The column height of each bin represents the surface area of the pleural cavity receiving a dose greater than or equal to that specific dose.

DISCUSSION

Table 1 shows the surface areas range from 886 cm² to 1716 cm² with a mean value of 1405±284 cm², and the volumes range from 2742 cm³ to 8192 cm³ with a mean value of 6704±1609 cm³. The scatter component, b, ranges from 6.5 mW/cm² to 7.8 mW/cm², with an average of 7.3±0.4 mW/cm², which suggests a slight variation.

Throughout the pleural PDT, the light source for light delivery was tracked, and the position data was recorded. Based on the stored data, the pleural cavity geometry was reconstructed (Fig. 1), and the light fluence distribution was calculated for the whole pleural cavity surface (Fig. 7). Together with the isotropic detector data, the detector positions were extrapolated and compared to the measured positions with the shifts calculated, as shown in Table 2. For case 004, there were results for six detectors instead of all eight because of the absence of position data for two detectors (AS and PS). This was one of the earliest cases, and it was challenging to treat these two areas without blocking of camera’s view due to inadequate experience in camera placement. This procedure was enhanced with more experience and knowledge about operation room arrangement in the following cases. The average difference between calculated and measured positions for each detector ranged from 2 to 3 cm. The maximum and minimum differences were 6.11 cm for case 017 and 0.04 cm for case 004, respectively. In general, the calculated and measured positions agree with each other, but both methods have certain limitations. Depending on the shape of the pleural cavity and camera position, the difficulty in manually measuring the detector position accurately varied. The detector position could shift slightly during the treatment after the measurement has been taken due to the patient’s body relocation. On the other side, the accuracy of the automatic calculation algorithm relies on efficient position data collection. The calculated results may contain a large bias if there is insufficient position data for a particular detector. The accuracy of the detector positions can be further validated by comparing calculated and measured light fluence data at these locations.

As shown in Fig. 3, the light fluence calculation only involved the primary component, which resulted in a relatively large disagreement at the end of the treatment. For each detector position, the sharply increased regions (e.g., the regions highlighted in grey in Fig. 3(h)) represent the features of both measured and calculated light fluences, meaning the area was treated and illuminated. These increments contribute to the total light fluence at the end of the treatment. The features on the measured and calculated fluence are expected to match up, which can be served as a sign of a valid light fluence calculation method, and, more importantly in this study, the precise detector positions used. When adding in the constant scattered component, the overall agreement was improved, as seen in Fig. 4. Figure 5 presents the calculated light fluence using the same methods (primary & scattered) but with the measured detector positions for comparison with Fig. 4 to evaluate how well the fluence matched. The figures demonstrate that the features (sharp increases in light fluence) match better with the detector positions computed by the newly developed algorithm, suggesting that the algorithm increased the overall agreement between calculated and measured light fluence. Figure 6 shows a further improvement in the light fluence calculation method by employing the dual correction factor (CF). The light fluence and detector position calculation were evaluated by comparing the calculated light fluence at the calculated locations of the eight isotropic detectors with the measured light fluence as shown by Figs. 3, 4, and 6 and summarized statistically in Table 3.

The percent error using the primary component (ϕ_primary) only for each case and detector position is summarized in Table 3(a). As shown in Fig. 3, the difference between the calculated and the measured light fluence is relatively large. Among all cases and detector positions, the maximum percentage error is 61.9% for case 038 at PCW, and the overall average for all cases is 49.1%±8.8%. The agreement is greatly improved by using a primary component with a constant scattered component, as shown in Fig. 4. For case 037 (Fig. 4), the average disagreement is 6.2%±4.0%, and can be seen that the features of measured and calculated fluence lines harmonize well. In comparison, Figure 5 presents results for the same cases using the same calculation methods but with the measured detector position, which results in a slightly higher disagreement of 10.3%±5.7%. And also, the features during treatment do not harmonize as well as the result shown in Fig. 4. Among all cases and detector positions, the maximum percentage error is 13.6% for case 017 at Diaph, and the overall average for all cases is 7.0%±3.4%. There is a slight improvement compared to the previous method based on the measured detector position, where the maximum error is 15.4% (15). The result for each case and detector position is summarized in Table 3(b). Inspired by the small variation in scatter component values across all cases, the average b value (7.3 mW/cm²) was used as a fixed constant scattered component for all cases, instead of a specific value for each case, to simplify the calculation of light fluence. The result is summarized in Table 3(c). A constant power value (5250 mW) of the light source is also used for all cases. The maximum percentage error increases to 20.9%, and the overall average increases to 8.4%±4.7%. Despite a slight increase in disagreement, the variation is acceptably low, which suggests that using a fixed average scattered component for light fluence calculation in pleural PDT is feasible.

Table 3(d) summarizes the result for calculation with primary component only and dual CF applied. The maximum percentage disagreement among all cases is 33.8% from case 014 at AS. The mean value for all detectors and cases is 12.7% ± 7.2%. This method is able to improve the agreement compared to using the primary component only, but the resulting errors are larger than the result from the calculation using primary and scattered component without dual CF applied. This conclusion holds even when the fixed scattered value is used. The agreement is improved the most by applying the dual CF to the calculation with both primary and scattered components (Fig. 6 and Table 3(e)). The maximum percent error among all cases is 11.9% for case 017at PM. The average across all detectors and cases is 5.5% ± 2.9%.

Figure 7 shows the light fluence distribution at the end of PDT treatment of 8 patients for whom light delivery was tracked for the entirety of PDT treatment (until 60 J/cm²) using the primary and scatter dose plus dual correction. The symbol (“x”) marked the location of the isotropic detectors. These locations vary for different patients because their coordinations are based on camera coordinations, which are positioned slightly different relative to patient anatomy. Since the treatment is based on isotropic detectors reaching 60 J/cm², the actual delivery of light fluence is not uniform.

To evaluate the light fluence uniformity delivered to the patient, the standard deviation of the mean profile along depth for different angles (first column of Table 4) and the standard deviation of the mean profile (second column of Table 4). The standard deviation refers to the percentage standard deviation from the ideal dose prescription, and the method for quantifying the uniformity is described in detail in the previous study (15, 16). The uniformities ranged from 4.6 % to 14.2%. Across all cases, the average uniformity is 9.3%, with a variation of the standard deviation of 19.1%. The resulting uniformity is comparable to the previous study (10%) (15).

To evaluate the minimum light fluence delivered to the patient, fluence-surface-histogram (FSH) are compared among patients for the treatment target (Fig. 8). For the pleural surface target, 80% of the total surface area was receiving a minimum of 50 J/cm² (83.3% of 60 J/cm²) for all eight patients evaluated (corresponding to fluence distribution in Fig. 7). Unlike radiation therapy, PDT is thought to have larger tolerance to the prescribed fluence, and an accuracy of 15% (vs. 5% for RT) is considered acceptable (17, 18).

CONCLUSIONS

As a standard of care for pleural PDT at the University of Pennsylvania, the light fluence is monitored using eight isotropic detectors. The current procedure requires pre-determination of the discrete detector positions. An automatic algorithm has been developed to be used with the optical IR navigation system to provide 2D light fluence distribution for the entire pleural surface without pre-treatment measurement of isotropic detector positions. This algorithm is validated by comparing the measured and calculated light fluence based on the extrapolated detector positions. Light fluence calculation using both the primary and scattered components is able to achieve agreement within 14% of the measured values for each detector for eleven clinical cases. With the primary component together with fixed power value and scatter component for all cases, the maximum percentage difference increases to around 20%, which is larger than those using dual CF but is still within an acceptable range. Fluence surface histogram (FSH) has been used to evaluate the minimum coverage and variation of the PDT treatment.