Comprehensive reanalysis of light fluence distribution in pleural photodynamic therapy using standardized anatomical coordinates

Abstract

Photodynamic therapy (PDT) has shown promise as an adjuvant treatment for malignant pleural mesothelioma when combined with surgical resection. Accurate light dosimetry is critical for treatment efficacy. This study presents an improved method for analyzing light fluence distribution in pleural PDT using a standardized anatomical coordinate system and advanced computational modeling. We utilized an infrared navigation system with an improved treatment delivery wand to track light delivery in real-time. The pleural cavity geometry was reconstructed and mapped to a standardized coordinate system, allowing for direct comparisons across patients. Light fluence was calculated using both primary and scattered components, with a novel dual correction method applied to match measured values at detector locations. The standardized approach allowed for statistical analysis of light fluence distribution across anatomical regions in a cohort of 11 patients. Results showed acceptable light fluence uniformity with a standard deviation of 6.6% from the prescribed dose across patients. This comprehensive analysis provides insights for optimizing treatment protocols and lays the groundwork for future studies on singlet oxygen generation and its correlation with treatment outcomes in pleural PDT.

Graphical Abstract

This study utilized an infrared navigation system to track light delivery in real-time in pleural PDT patient treatment. The pleural cavity geometry was reconstructed and mapped to a patient coordinate system, allowing for direct comparisons across patients. A comprehensive statistical analysis of light fluence distribution (based on both calculation on the entire pleural cavity and measurements in 8 sites per patient) were performed in a cohort of 11 patients. Results showed an average 86.9% (± 6.3%) of the pleural surface area received 90% of prescribed light dose (60 J/cm²) across patients.

INTRODUCTION

Malignant pleural mesothelioma (MPM) is an aggressive cancer with a poor prognosis. Multimodal treatment approaches, including surgical resection combined with intraoperative photodynamic therapy (PDT), have shown promise in improving local tumor control and overall survival (1-4). PDT involves the administration of a photosensitizing agent followed by light activation, leading to the generation of cytotoxic singlet oxygen species (5-7).

In the data analysis in our previous paper (8), camera coordinates were used resulting the position of each detector to be inconsistent because the position of the camera relative to the patient changes every time. The detector positions for each site should be roughly the same relative to the patient anatomy. In the study, we have reanalyzed light fluence rate data for all these patients using the patient coordinates, thus one can judge whether the detector position for eight sites (Apex, PCW, ACW, PS, AS, PM, Peri, Diaph) are reasonable or not. The light fluence distribution in Figure 6, is now aligned to be consistent with patient anatomy, i.e., the depth direction (z-axis) is now always along the patient superior inferior direction. In addition, light fluence rate distribution data was completely reanalyzed with correction to include dual field correction rather than primary plus scatter without dual field correction so that the resulting light fluence distribution is much more accurate to reflect the actual light delivery.

Figure 6.

Light fluence distribution maps on the inner surface of standardized pleural cavities for 11 cases. The use of a consistent coordinate system facilitates inter-case comparison and identification of frequently under-treated regions. Detector positions are indicated by black crosses, demonstrating their relatively fixed locations across cases.

Accurate light dosimetry is crucial for PDT efficacy, as both under- and over-treatment can lead to suboptimal outcomes (9, 10). Previous studies have utilized discrete isotropic detectors to monitor light fluence at specific locations within the pleural cavity (11, 12). While this approach accurately measures the light fluence at discrete locations, it does not provide information about 2D light distribution throughout the entire treatment surface (13).

Recent advancements in optical tracking technologies and computational modeling have opened new avenues for improving PDT dosimetry (14). In this study, we present a novel approach that combines tracking of light delivery, standardized anatomical mapping of the pleural cavity, and advanced computational modeling to provide a comprehensive analysis of light fluence distribution in pleural PDT. This improved understanding of light distribution lays the foundation for future work on modeling singlet oxygen generation and optimizing treatment protocols.

MATERIALS AND METHODS

Patient Selection and Treatment Protocol:

Eleven patients with MPM were enrolled in a clinical trial (ClinicalTrials.gov Identifier: NCT02153229) at our institution. All patients underwent surgical resection followed by intraoperative PDT. The photosensitizer Photofrin was administered at a dose of 2 mg/kg body weight 24 hours prior to surgery. Illumination was performed using a 630 nm wavelength diode laser (Modulight, Tampere, Finland) to deliver a prescribed light fluence of 60 J/cm² (15).

Improved Treatment Delivery Wand and Infrared Navigation System:

An improved treatment delivery wand that allows for direct tracking of the fiber tip position has been developed and employed in patient treatments. The wand consists of a modified endotracheal tube with a balloon applicator tip, housing the optical fiber. Nine passive reflective markers are attached to the wand, allowing for real-time tracking by an infrared camera system (Polaris Spectra, NDI, Waterloo, Canada).

The infrared navigation system tracks the position and orientation of the treatment wand at a rate of 20-60 Hz, providing continuous 3D coordinates of the light source during treatment (Figure 1). The accuracy of the system is approximately 0.5 mm in 3D, with a maximum detection volume of 205 × 186 × 147 cm³ (16).

Figure 1.

Image of the treatment wand and IR camera in use during pleural PDT. The wand, controlled by physicians, delivers light uniformly to the inner surface of the pleural cavity. The camera, positioned above the patient, monitors the light source position throughout the treatment.

Pleural Cavity Geometry Reconstruction and Standardization:

For PDT treatment, we defined eight essential anatomical locations that serve as key reference points for treatment planning and analysis (Figure 2). The apex, located at the uppermost point of the pleural cavity, typically represents the superior extent of the lung. The posterior chest wall (PCW) is situated along the back of the chest cavity, adjacent to the vertebral column, while the anterior chest wall (ACW) is located at the front of the chest cavity, near the sternum. The posterior sulcus (PS) and anterior sulcus (AS) represent the posterior-superior and anterior-superior corners of the pleural space, respectively. The posterior mediastinum (PM) refers to the region adjacent to the heart and major blood vessels on the posterior side of the chest cavity. The pericardium (Peri) area surrounds the heart and is critical for ensuring adequate treatment while minimizing risk to the cardiac tissue. Finally, the diaphragm (Diaph) forms the inferior boundary of the pleural cavity, separating the thoracic and abdominal cavities.

Figure 2.

Schematic representation of the eight standard detector locations in the pleural cavity for PDT treatment. The 3D volume depicted is a reconstructed pleural cavity (left lung) based on patient CT data. Detector positions: apex, posterior chest wall (PCW), anterior chest wall (ACW), posterior sulcus (PS), anterior sulcus (AS), posterior mediastinum (PM), pericardium (Peri), and diaphragm (Diaph).

For every case, the pleural cavity geometry was reconstructed using a novel algorithm that processes the position data obtained from the infrared tracking system. This algorithm employs a spherical coordinate transformation approach to efficiently handle the complex topology of the pleural space. Detailed illustration can be found elsewhere (17). The contour, initially reconstructed directly from the raw position data, was represented in the camera's intrinsic coordinate system. However, this coordinate system was subject to change depending on the camera's positioning, which could vary between different treatment sessions. This variability in the camera's coordinate system posed a challenge for consistent and accurate contour representation across multiple treatments.

To enable meaningful comparisons across patients, we developed a standardized anatomical coordinate system. This system orients the reconstructed cavity geometry such that the apex is aligned with the positive z-axis. The standardization process involves two key steps. First, we establish a "standard coordinate" by recording a reference position with the treatment wand held parallel to the patient's body, with the reflective markers pointing towards the head (Figure 3). Second, we apply a rotation matrix to align the cavity geometry with this standard orientation. These anatomical locations were consistently identified across all patients using our standardized coordinate system. Their positions were determined through a combination of automated geometric analysis and manual verification by experienced clinicians.

Figure 3.

Demonstration of the "standard coordinate" establishment process. The treatment wand is positioned parallel to the supine patient's body, with reflective markers oriented toward the head. This reference position enables consistent orientation of pleural cavity reconstructions across patients. The coordinates are defined as the following: z for superior-inferior direction, y for right-left directions, and x for anterior-posterior direction.

To facilitate visualization and analysis, the 3D cavity geometry is "unwrapped" into a 2D representation for light distribution evaluation. In this format, the horizontal axis represents the circumferential angle around the cavity, while the vertical axis corresponds to the superior-inferior position. This unwrapped view allows for easier comparison of treatment parameters across different anatomical regions and between patients.

The standardized geometry and consistent identification of key anatomical locations enable several important analyses. We can now quantitatively compare cavity shapes and sizes across the patient cohort and evaluate light fluence distribution in relation to specific anatomical features. Furthermore, this approach allows for statistical analysis of treatment parameters at corresponding locations across multiple patients, facilitating the development of region-specific treatment guidelines based on aggregate data from the entire cohort.

Light Fluence Calculations:

Light fluence was calculated as the sum of primary (direct) and scattered components:

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

where $\mathrm{S}$ is the source power, $\mathrm{r}$ is the distance from the light source to the point of interest, $\mathrm{b}$ is a constant scattered light contribution, and $\text{CF}(\mathrm{t})$ is a time-dependent dual correction factor.

The dual correction method was implemented to match calculated fluence values to measured values at detector locations. This approach applies a time-dependent multiplication factor to the entire calculated fluence distribution, improving agreement between calculated and measured fluence at detector points to within 12% for most cases (8).

RESULTS

Raw Pleural Cavity Reconstructions

Figure 4 presents the reconstructed pleural cavities for all 11 cases using the raw position data obtained from the infrared tracking system. These reconstructions are shown in the original camera coordinate system, which varies for each treatment session due to differences in camera positioning. The variability in orientation and shape among these raw reconstructions highlights the challenge of direct inter-patient comparisons without standardization.

Figure 4.

Reconstructed pleural cavities for 11 cases using raw position data, shown in the original camera coordinate system. Note the varying orientations due to differences in camera positioning during each treatment session.

Standardized Pleural Cavity Representations

Following the application of our standardization algorithm, Figure 5 displays the same 11 pleural cavities oriented in the "standard coordinate" system. In this representation, the apex of each cavity is consistently positioned at the top of the image. The coordinate system is defined with the z-axis oriented along the inferior-superior direction, the y-axis representing the left-right direction, and the x-axis aligned with the anterior-posterior direction. While the essential anatomical locations may not appear exactly fixed due to inter-patient variability, cavity distortions, and differences between left and right-sided treatments, their relative positions remain consistent across cases. The standardized representations also enable determination of treatment laterality through the relative positions of the detectors. As shown in Figure 5, the spatial relationships clearly identify the treated cavities as: left lung (04, 08, 12, 16, 17, 18, 20, 38) and right lung (14, 27, 37). This spatial pattern of detector positions serves as an anatomical reference that can verify treatment side independently of surgical records, enhancing the reliability of our treatment documentation and analysis.

Figure 5.

Standardized representations of reconstructed pleural cavities for 11 cases. Volumes are oriented in the "standard coordinate" system with the apex consistently positioned at the top. While essential anatomical locations may not appear exactly fixed due to inter-patient variability and cavity distortions, their relative positions remain consistent. Side of lungs treated: left lung (04), left lung (08), left lung (12), right lung (14), left lung (16), left lung (17), left lung (18), left lung (20), right lung (27), right lung (37), and left lung (38).

During treatment, only the reference orientation along the z-axis (inferior-superior direction) was acquired. This single-axis reference ensures the validity of the inferior-superior orientation in the standardized coordinate system. The orientations along the remaining axes were determined post-treatment through examination of relative detector positions and manual adjustments to achieve anatomically correct alignment. This approach, while not fully automated, provides sufficient anatomical consistency for meaningful inter-patient comparisons. The standardized representations reveal similarities in overall cavity shape across patients, with notable variations observed in the curvature of the posterior wall and the depth of the costophrenic recess. These variations underscore the importance of patient-specific treatment planning even within a standardized framework.

Light Fluence Distribution

Figure 6 illustrates the light fluence distribution maps on the inner surface of the standardized pleural cavities for all 11 cases. The use of a consistent coordinate system facilitates inter-case comparison and enables the identification of regions that consistently receive higher or lower light doses. The detector positions, marked consistently across all cases, exhibit spatial coherence in the standardized coordinate system, facilitating robust comparative analysis. The integration of standardized coordinates with the dual correction method yields light distribution maps that better reflect actual clinical conditions, demonstrating improved spatial consistency of detector positions and more uniform cumulative light distribution compared to previous studies. This enhanced representation provides a more accurate visualization of the treatment delivery patterns within the pleural cavity (8, 16, 17).

The comparison between calculated and measured light doses across all detector positions is summarized in Table 1. The results showed good agreement, with most variations falling within ±10%. The overall average difference across all detectors and cases was minimal, with detector-specific averages ranging from −3.9% (Apex) to 3.6% (ACW). Individual variations were well-controlled, with standard deviations typically between 4.6% and 6.4% across different detector positions.

Table 1.

Summary of the percentage difference doses (%) for each detector between calculated and measured light dose (${\varphi }_{\text{cal}}∕{\varphi }_{\text{meas}}-1$), for all available cases at the end of the treatment.

Case no.	PM (%)	Apex (%)	Peri (%)	ACW (%)	AS (%)	PCW (%)	Diaph (%)	PS (%)	Average (%)
04	−3.3	−9.3	−5.3	10.3	1.4	−1.6	−10.5	6.7	−1.5 ± 7.3
08	5.3	3.1	−5.8	−3.3	4.0	3.4	6.8	7.7	2.7 ± 4.8
12	6.3	1.5	−1.3	8.1	9.2	−3.3	−4.5	−5.2	1.4 ± 5.8
14	0.1	−3.6	−6.6	−0.5	5.4	3.2	− 0.3	2.8	0.1 ± 3.9
16	−2.4	9.4	−6.9	6.3	−3.1	3.6	1.8	−8.5	0.0 ± 6.3
17	−0.4	−1.0	3.2	2.2	11.1	4.3	−10.6	−4.8	0.5 ± 6.5
18	3.3	−6.3	4.4	5.0	8.5	−1.4	7.6	−0.1	2.6 ± 5.0
20	0.8	−8.5	−7.5	−6.6	−10.4	8.7	4.7	5.0	−1.7 ± 7.4
27	−9.8	−10.2	11.1	6.8	−1.3	−6.5	−0.7	0.5	−1.3 ± 7.6
37	1.1	−9.5	4.3	3.2	1.4	−7.9	3.4	−2.8	−0.9 ± 6.3
38	−4.6	−8.0	−5.7	7.9	8.3	8.0	2.1	10.1	2.3 ± 7.4
Average (%)	−0.3 ± 4.6	−3.9 ± 6.4	−1.5 ± 6.3	3.6 ± 5.2	3.1 ± 6.4	1.0 ± 5.5	0.0 ± 6.2	1.0 ± 6.0

Light fluence profiles were analyzed as a function of horizontal angle at each depth z for all patients (Figure 7). The mean fluence is represented by a solid black line, with the corresponding standard deviation shown as a grey-shaded region, referenced against the prescribed dose of 60 J cm⁻² (dashed line). Region between the two vertical dotted black lines delineate the estimated locations of surgical openings. Notable peaks in fluence appear at specific angles, corresponding to these surgical access regions, arising from the accumulated position data recorded during instrument insertion and withdrawal through the surgical cavity. These elevated readings reflect the increased sampling of position data at the surgical entry points rather than actual treatment delivery patterns.

Figure 7.

Fluence distribution along the horizontal angle for each depth (z-axis) for all available cases. The mean is shown in a solid black line, and the grey area indicates the standard deviation. The horizontal dashed line indicates the prescription light dose of 60 Jcm⁻². The vertical dotted lines outline the region corresponding to the surgical opening.

Fluence-surface-histograms (FSH) were generated to evaluate the minimum light fluence delivery across the treatment target (Figure 8). The fluence calculations excluded data from regions identified as surgical openings, as indicated in Figure 7, to prevent artificial elevation of values due to instrument transit artifacts. The uniformity of light fluence distribution and surface coverage analysis is presented in Table 2. Analysis revealed an average standard deviation of 6.6 ± 1.9% from the prescribed dose, with case-specific variations ranging from 4.2% to 10.4%. Analysis of the pleural surface coverage demonstrated that for all eight evaluated patients, 86.7% of the total surface area received a minimum light fluence of 54 J cm⁻², representing 90% of the prescribed dose (60 J cm⁻²).

Figure 8.

Fluence distribution along the z-axis (depth) for each angular location for all available cases. The mean is shown in a solid black line, and the grey area indicates the standard deviation. The horizontal dashed line indicates the prescription light dose of 60 Jcm⁻². The vertical dotted lines outline the region corresponding to the surgical opening.

Table 2.

Summary of uniformity for profiles in figure 7 (excluding the region corresponding to the surgical opening, outlined by the vertical dotted line), the percentage surface coverage for 54 J/cm², and the percentage surface coverage for 60 J/cm².

Case no.	Standard deviation (% difference)	Variation of std. deviation (% difference)	FSH90% (%)#	FSH100% (%) *
04	4.2	5.7	82.9	51.7
08	6.7	12.9	96.2	72.5
12	4.6	5.9	81.0	53.9
14	6.8	13.4	93.4	54.3
16	5.3	8.4	85.8	48.5
17	4.7	13.5	90.6	54.9
18	8.1	13.6	79.3	56.6
20	7.9	10.8	86.5	61.2
27	10.4	13.9	79.6	49.4
37	5.9	5.6	83.9	49.0
38	7.9	7.1	96.2	72.8
Average	6.6 ± 1.9	10.1 ± 3.6	86.9 ± 6.3	56.8 ± 8.7

^#Surface coverage for 54 J/cm²

^*Surface coverage for 60 J/cm²

DISCUSSION

The progression from raw pleural cavity reconstructions (Figure 4) to standardized representations (Figure 5) demonstrates the power of our approach in enabling meaningful inter-patient comparisons. The variability observed in the raw reconstructions underscores the necessity of a standardized coordinate system for consistent analysis across treatments and patients.

The standardized cavity representations (Figure 5) reveal important insights into the anatomical variability of the pleural space across patients. While overall cavity shapes show remarkable similarities, the observed variations in posterior wall curvature and costophrenic recess depth highlight the need for individualized treatment planning. These patient-specific variations could significantly impact light distribution and, consequently, treatment efficacy. Furthermore, the standardized coordinate system enables systematic verification of isotropic detector positions, which were previously recorded through manual measurements during surgery. The standardized representation allows visualization of relative detector positions, providing insights into the laterality of the treated lung and enhancing our understanding of patient-specific anatomical conditions. This spatial relationship among detector locations serves as an additional anatomical reference, enabling retrospective confirmation of treatment side and anatomical orientation without relying solely on surgical records. While our previous work established an algorithm for calculating detector positions using navigation and light fluence data (8), significantly enhances the accuracy of detector position determination. While the earlier algorithm provided a computational approach to determine detector locations using navigation and light fluence data, the addition of standardized cavity representation now allows for verification and correction of detector identification. This complementary approach addresses both positional and nomenclature uncertainties, resulting in more reliable detector position documentation for treatment analysis.

The reanalysis of light fluence distribution in this study reveals significantly more uniform treatment delivery patterns compared to our previous work (8). A critical improvement stems from the correction of previously mislabeled detector names, which was made possible through our standardized anatomical coordinate system. By rectifying these detector identifications and applying the dual correction method, the light fluence distribution maps (Figure 6) now show markedly improved uniformity and more logical spatial patterns than previously reported. In particular, the light distribution now follows expected patterns of treatment delivery, with gradual transitions between adjacent anatomical regions rather than the abrupt variations seen in our earlier analysis.

This enhanced uniformity makes anatomical sense, as the corrected detector positions now properly reflect the natural contours of the pleural cavity. For instance, the eight key anatomical locations (Apex, PCW, ACW, PS, AS, PM, Peri, Diaph) show spatially coherent light fluence patterns that align with the physical constraints of light delivery during treatment. The dual correction method further refines these distributions by accounting for both primary and scattered light components, resulting in fluence maps that better represent the actual treatment delivery patterns within the pleural cavity. These improvements provide a more reliable foundation for evaluating treatment coverage and optimizing light delivery strategies.

The light fluence distribution maps (Figure 6) provide a comprehensive view of treatment delivery across the entire pleural surface, addressing a key limitation of discrete detector measurements. The consistency in detector positioning across standardized cavities allows for more reliable comparisons between cases and identification of areas that may consistently receive higher or lower light doses.

While this study focused on light fluence distribution, the standardized framework we've established lays the groundwork for future investigations into singlet oxygen generation. By incorporating models of photosensitizer distribution and tissue oxygenation into our standardized cavity representations, we could develop a more comprehensive understanding of PDT efficacy throughout the pleural space (18-20).

A promising avenue for future research involves importing our corrected and standardized pleural cavity contours into advanced modeling software such as COMSOL Multiphysics (21). This approach offers several advantages for enhancing our understanding and optimization of pleural PDT. This comprehensive modeling approach could provide valuable insights into spatial variations of treatment efficacy and help identify areas at risk of under-treatment due to factors beyond light delivery alone (22-24). While implementing such advanced modeling presents technical challenges and would require careful validation, it represents a significant step towards personalized, highly effective PDT treatments for malignant pleural mesothelioma.

Furthermore, the ability to consistently map treatment parameters across a patient cohort opens new avenues for correlative studies. Future research could investigate relationships between light fluence patterns in specific anatomical regions and treatment outcomes, potentially leading to the development of region-specific treatment guidelines.

The standardized anatomical coordinate system we've developed could readily accommodate the implementation of different photosensitizers with varying efficacies. While our current study focused on Photofrin-mediated PDT, the framework could be adapted for other photosensitizers by incorporating their specific optical properties, quantum yields, and activation wavelengths into the light fluence calculations. The dual correction method could be particularly valuable for photosensitizers with different scatter and absorption characteristics, as it can account for variations in light propagation through tissue. Additionally, for photosensitizers with different therapeutic windows or depth penetration properties, our standardized mapping approach could help identify optimal light delivery strategies and treatment parameters specific to each agent. This flexibility makes our framework particularly valuable for evaluating and comparing novel photosensitizers in future clinical trials.

CONCLUSION

The standardized anatomical mapping approach, combined with improved light delivery tracking and computational modeling, provides a powerful framework for comprehensive analysis of light fluence distribution in pleural PDT treatments. This method facilitates clearer visualization of light fluence throughout the treatment volume, potentially leading to optimized treatment protocols. Future work building on this foundation to model singlet oxygen generation could further enhance our understanding of treatment efficacy and guide improvements in pleural PDT for patients with malignant pleural mesothelioma.

Figure 9.

Cumulative light fluence-surface histograms (FSH) for eight patients in standardized coordinates using light fluence obtained using primary & scatter components with dual CF. Each histogram bin height represents the percentage of pleural cavity surface area receiving a light fluence greater than or equal to the corresponding fluence value. This visualization enables quantitative comparison of light dose coverage across patients.