Open Access
Add to BibSonomy
Abstract
Photodynamic therapy (PDT) has shown significant potential for skin disease treatment. As a key element, light is critical to influencing its treatment outcome, and light dosimetry is an issue of much concern for researchers. However, because of three-dimensional irregularity in shape and patient’s movement during the therapy, irradiance hardly keeps uniform on the lesion and flux measurement remains a challenge. In this work, we report the development of a three-dimensional image-guided PDT system, and the method of dynamic irradiance planning and flux monitoring for lesions in different poses. This system comprises a three-dimensional camera for monitoring patients’ movement during therapy, a computer for data analysis and processing, and a homemade LED array for forming uniform irradiance on lesions. Simulations on lesions of the face and arm show that the proposed system significantly increases effective therapy area, enhances irradiance uniformity, is able to visualize flux on the lesion, and reduces risks of burns during PDT. The developed PDT system is promising for optimizing procedures of PDT and providing better treatment outcomes by delivering controllable irradiance and flux on lesions even when a patient is moving.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Patients with skin diseases, such as vitiligo, metastatic melanoma and port-wine stains, are tortured psychologically and physically [1–3]. It has been reported that approximately more than 63 million people are newly affected by vitiligo (0.5–2%), metastatic melanoma (1.9%) and port-wine stains (0.3%–0.5%) each year according to their prevalence measurements [4–7]. Photodynamic therapy (PDT) is currently a promising method for treating different skin diseases [8,9]. PDT has three essential elements, oxygen, photosensitizers and light. Combined with oxygen provided by patients, light with proper wavelength and intensity activates injected photosensitizers to generate reactive oxygen species (ROS), which affect all intracellular components, including proteins and DNA, resulting in necrosis or apoptosis to achieve the purpose of treatment [10,11]. Due to the advantages of noninvasiveness and low adverse effects, PDT has gradually become a globally accepted method in clinical studies [12].
As an essential element, light plays an important role in PDT. In clinical settings, PDT is generally prescribed by the irradiance (mW/cm2) and light flux (J/cm2) for the treatment device, while the effective light dosimetry on the lesion remains a challenge [13]. Appropriate light flux could induce target tissue apoptosis, but overdosing light flux would cause burns and scars [14,15]. Uniform and controllable irradiance distribution on the lesion and light flux monitoring are necessary for PDT treatment [16]. Human skin shows a three-dimensional complex structure, especially the facial region, which results in a conventional plane therapeutic light source forming an un-uniform spot on the skin lesion [17]. With the accumulation of treatment time, there is a lack of information on the light flux distribution of the lesion. Thus, doctors have to adopt lower irradiance to avoid topical burns, which results in a portion of the skin receiving insufficient light flux therefore patients need repeated treatment after several months [18]. Many studies report novel PDT equipment and devices to improve irradiance uniformity on the skin lesion. A “light blanket” has also been proposed, in which a diffusing optical fibre is wrapped inside a flexible plastic layer to provide uniform irradiance [19]. A couple of textile diffusers have also been developed with plastic optical fibers [20,21]. These can be used to cover the lesion site during irradiation with their flexibility. Despite these studies have demonstrated that the fiber textiles could improve the irradiance uniformity of the field, the irradiance is too low (maximum 20 mW/cm2) to achieve the therapeutic purpose for most treatments [22]. For a laser fiber with limited illumination area, a sufficient light flux and uniform irradiance can be obtained in such a way that the surgeon has to constantly move the fiber so that the center of the laser spot is kept in motion over the lesion [23]. An adjustable LED device can be applied in different areas of the body [24], ensuring uniform light distribution and reducing the time of the operation, but the discomfort and cross-infection caused by its attachment to the skin lesions must be considered.
Another problem also plagues clinicians. PDT usually lasts for tens of minutes [25], and patients may move due to PDT induced-pain [26], medical staff needs to adjust the position of the light source during the surgery, resulting in physical burden. This long-duration manual operation is also a high workload for doctors and is very reliant on experience with low efficiency and accuracy. Uniform illumination using an array of LEDs provides a novel solution for the problems mentioned above [27]. A large number of studies have demonstrated that a good uniform irradiance on the projection surface can be achieved by optimizing LED array parameters, such as LED power allocation [28], LED normal vector orientation [29], and LED arrangement [30], etc. In this way, the LED array-based light source can perform illumination flexibility when the relative position relationship between the LED array and the target point is known [31].
Therefore, we propose a three-dimensional image-guided photodynamic therapy system for skin disease surgery in this study. Three-dimensional visual sensors have been widely used to be one of the best options for the acquisition of images with distance information, and are used to obtain the relative position and rotation relationship between the LED array and the lesion in the proposed PDT system. The system has the ability to form a more uniform irradiance on the complex three-dimensional lesion surface and adjust the LED-based light source according to the pose of the lesion, addressing the problems of irradiance uniformity, light flux monitoring, and intra-operative patients moving.
2. Methods and materials
2.1 System setup
As shown in Fig. 1(a), a three-dimensional image-guided photodynamic therapy system with irradiance planning for skin diseases has been developed, which consists of a computer (Redmi G 2021, Xiaomi, China), a homemade LED array, and a three-dimensional camera (Surface HD 50, Revopoint, China) with a resolution of depth map of 1290 × 1200, resolution of color image of 1290 × 1080, and field of view of H55°× V36°. In order to observe patients as widely as possible, any part of the LED array does not show up in this field of view. The point accuracy of the 3D camera is ±0.15 mm at a distance of 50 cm. The LED array comprises 14 rows and 7 columns of LEDs with an interval of 1 cm. The specifications of each LED (XLamp XQ-E LEDs, Cree, USA) are the same, the brightness of each LED can be adjusted by pulse width modulation (PWM) with the irradiance intensity up to 750 mW/Sr. The lens (FP16558_LISA3-RS-PIN, Ledil, Finland) coupled efficiency is 90%, with a half angle of 15 deg. Codes are executed on a computer with 8 GB RAM. The processor is an 11th Gen Intel Core i5-11260 H CPU. Figure 1(b) shows the physical photograph of the system, and Fig. 1(c) illustrates the workflow. After the image acquisition of the patient, the 3D camera delivers the whole image in its field of view to the computer, then the computer executes the following steps to achieve uniform irradiance on the lesion. First, pose calculation: the pose relationship between the lesion and the LED array must be solved to calculate the brightness matrix precisely. Second, image segmentation: the lesion portion is segmented from the whole point cloud image according to the color difference between the normal skin and the lesion. Next, a corrected LED array illumination model is established to find the correspondence between the brightness matrix and the lesion. Finally, solve and pass the optimal brightness matrix to the LED array.
Fig. 1. System composition and workflow. (a) The scheme of the proposed system, includes a computer, a three-dimensional camera and a homemade LED array. The LED array can form a uniform irradiance on a lesion according to its shape and size. (b) Physical image of the proposed system. (c) The workflow of the system.
2.2 Pose transformation
Cam 1 has the responsibility to monitor a patient’s movement, however as its coordinate system does not contain any information about the LED array, we introduce another three-dimensional camera (Cam 2) and a reference object, as shown in Fig. 2(a), to translate the position of the LED array into the coordinate system of Cam 1. The Cam 2 (Surface120, Revopoint, China) has a resolution of depth map of 640 × 400, a resolution of color image of 2560 × 1600, and a field of view of H52° × V31°. The point accuracy of the 3D scanner is ± 0.22 mm at a distance of 50 cm. This reference object is a cube with a side length of 8 cm, each pair of opposite sides of the cube are painted with the same colour, while the adjacent sides are painted with different colours.
Fig. 2. Approach to pose calculation. (a) Scheme of pose calculation system, including a known size reference object, a cube here, and two 3D cameras (Cam 1 and Cam 2). (b) Registration of point cloud captured by Cam 1 and Cam 2. Images of the pose relationship between patient and LED array in physical world (c) and Cam 1 (d).
Cam 2 is placed parallelly at 80 cm away from the LED array. Move the reference object to the common field of view of these two cameras. When the two cameras acquire the point cloud data of the reference object at the same time, Cam 2 acquires the point cloud data of the LED array. Through a point cloud registration process for the reference object [32], the transformation matrix ${T_c}$ of the two cameras can be obtained, as described in Eq. (1), where ${c^1}$ represents the coordinate of Cam 1 and ${c^2}$ represents the coordinate of Cam 2. Using this transformation matrix, the LED array point cloud can be transformed into the coordinate system of Cam 1.
2.3 Effective irradiance mathematical modelling
To solve the optimal brightness matrix of the LED array for a certain surface, a LED array illumination mathematical model is essential to be established. The irradiance of a certain point on the lesion from a single LED source can be expressed as [30]:
$\theta $ is the angle between the normal vector at a point on the lesion and the connection vector from an LED to the point on the lesion, and r is the distance between this point and the LED, where the normal vector is obtained by calculating the triangle patch formed by adjacent three points.The LED can be roughly viewed as a Lambertian source [31], with the light intensity profile indicated by Eq. (3), where ${I_0}$ is the luminance intensity in the normal direction corresponding to the source surface and $\varphi $ is the divergence angle.
For an individually adjustable brightness LED, the ${I_0}$ can be described as:
The irradiance ${E_p}$ at a given point should be equal to the sum of the irradiance contribution to this point of all LEDs, as follows:
In addition, our previous study [34] has been adopted to correct the LED array illumination model due to lens installation errors and non-ideal LED light source. After correction, coefficients m and r of each LED are updated with correction coefficients ${c_m}$ and ${c_r}$ to each parameter respectively. The corrected LED array illumination model is described as follows:
2.4 Optimal brightness matrix
The irradiance of each point on the lesion point cloud is expected to be the target irradiance ${E_T}$, the optimization purpose is to find an optimal brightness matrix (14 rows and 7 columns), in which the brightness matrix is dependent on the $pwm$ matrix, to minimize the difference of ${E_p}$ and ${E_T}$ at each point. This process can be expressed as:
Mimicking the patient’s moving during PDT for two poses of the same lesion, the pose transformation of rotation and translation in Euclidean space can be described by a transformation matrix ${T_l}$ as:
2.5 Light dose monitoring
After obtaining the optimal brightness matrix, the irradiance of each point on the lesion can be calculated. According to the coordinate of a certain point at any time t, it can be inferred from Eq. (11) that the irradiance of the point can be expressed as ${E_p}(t )$. Flux (J/cm2) is the accumulation of irradiance (mW/cm2) and time (s) as:
3. Results
3.1 System composition and workflow
To achieve image guidance and irradiance uniformity dynamic optimization, a photodynamic therapy system is established, consisting of a three-dimensional camera (3D camera), a homemade LED array light source, and a computer, as shown in Fig. 1(a). The 3D camera is used to obtain the 3D points cloud of the patient and sends it to the computer, which calculates the optimization parameter of the LED array according to the lesion’s shape and size. Each LED’s brightness in this array can be adjusted individually (see Figure S1 in Supplement 1), hence, diverse brightness matrixes can form uniform irradiance on complex 3D surfaces.
3.2 Pose calculation and transformation
Due to the limited field of view of the camera, the LED array cannot be covered by the camera’s view (Cam 1) when the camera is monitoring the patient. Therefore, another 3D camera (Cam 2) and a reference object (a cube here) are introduced into the system temporarily to acquire the pose relationship between the LED array and Cam 1, as shown in Fig. 2(a). The reference object is placed within the common field of view of the two cameras, and each camera gains three adjacent surfaces of the reference object (marked as a blue box and a red box, respectively). At the same time, the point cloud data of the LED array has been established in the coordinate system of Cam 2. For a reference object with a cube shape corresponding to the colored opposite surfaces, a multi-view registration method is adopted (Sec.2.2), and the result is displayed in Fig. 2(b). The transform matrix ${T_c}$ of the coordinate system from Cam 2 to Cam 1 can be obtained after the registration process, as described below. The coordinate of the LED array is transformed to the coordinate system of Cam 1.
The registration step is no longer required in future surgical operations, as the position of the LED array is fixed in the coordinate system of Cam 1 and the transform matrix is invariant in this system. Figure 2(c) and Fig. 2(d) show the LED array and the patient in the physical world (${T_w}$) and Cam 1 coordinate system (${T_c}$), respectively.
3.3 LED array illumination model and lesion point cloud
To control the irradiance distribution, an accurate LED array illumination mathematical model must be established. An ideal LED light source could be regarded as a point. However, when soldering to a printed circuit board, LEDs will inevitably deviate from their positions with the tolerance of ±0.1∼±0.2 mm, and the lenses installation will further deviate from their ideal positions (see Figure S2 in Supplement 1), the outgoing direction of every ray is skewed. The light center of each LED deviated from its original position and the divergence angle is changed. Therefore, a regularly arranged LED array model as in the physical display (Fig. 2(c) and (d)) is no longer applicable in this system for the reason that the brightness matrix of the LED array has to be determined by precise positions of LEDs concerning the lesion point cloud. A corrected LED array illumination model has to be found to replace this LED array light source. A measuring method for the light center and divergence angle of the LED array and the corresponding uniformity calibration method has been proposed in our previous study [34], mainly establishing a LED array illumination model utilizing computer vision technology. In this study, we adopt the same measuring method, the system composition is shown in Fig. 3(a). When turning on the LEDs one by one, the corresponding light spot forming on a translucent board is recorded by Cam 2 (Fig. 3(b)). The contour of the spot can be easily depicted using an edge detection algorithm. Compared with a known-size reference coin, we can calculate the size of the contour (a detailed algorithm can be seen in Figure S3 in Supplement 1). After moving the plate in the direction perpendicular to the plane of the translucent board, the size of the light spot will change accordingly. The light center and divergence angle of each LED can be obtained through similar triangle principle (see Table S1 in Supplement 1), and the corrected LED array illumination model is described as Eq. (8). To validate the accuracy of the corrected LED array illumination model, actual measurements are carried out and compared with simulation values. As shown in Fig. 3(c), simulation values of different PWM values ranging from 16 to 240 for a LED are compared with the actual measurements of irradiance. Three LEDs are tested in such a way and their average values are taken. The result shows that the Pearson correlation coefficient is 0.9995, and demonstrates that the single LED illumination model is approximate to an actual LED. For the whole LED array illumination model, an optical power meter is used to measure the irradiance at the distance of 10 cm and 12 cm away from the LED array, respectively (see Figure S4 in Supplement 1). Similarly, simulation values are also calculated. Figure 3(d) shows irradiance distribution statistical results. In the same irradiated plane, the maximum irradiance, minimum irradiance and average irradiance of measurement value and simulation value are almost equal, Pearson correlation coefficient is 0.9986, which demonstrates that the whole LED array illumination model is approximate to the actual LED array. Therefore, the physical LED array is replaced by the corrected LED array illumination model (Fig. 3(e)). Supplementary Table S2 lists all coordinates of LEDs in Cam 1 coordinate system.
Fig. 3. LED array illumination model and correction. (a) Scheme of a measuring method for light center and divergence angle of LED array, including a translucent board, on which is used to LED spot image and two three-dimensional cameras (Cam 1 and Cam 2). (b) Physical image of the measuring method system. (c) Comparison of measurement values and simulation values of irradiance (same three LEDs in each experiment group). (d) Comparison of measurement values and simulation values of irradiance (whole LED array). (e) The pose relationship between patient and LED array in Cam 1. (f) Segmented lesion point cloud. (g) The pose relationship between the segmented lesion and LED array in Cam 1.
For the proposed system, a simple and effective lesion segmentation method is essential for the workflow. A color difference-based image segmentation algorithm in our previous study [35] is adopted, which can be used to extract the lesion point cloud from the whole patient image efficiently and accurately, and the segmentation result is shown in Fig. 3(f), the dimension of the bonding box of the lesion is 7.2 cm × 4.4 cm × 1.3 cm. Figure 3(g) shows the pose relationship between the corrected LED array illumination model and the lesion point cloud.
3.4 Irradiance distribution on lesions
In an LED array, each single LED has specific irradiance distribution, which has been obtained from the last section, and when turned on in a certain way with an appropriate brightness composition, the LED array can produce more uniform illumination. The brightness composition is a brightness matrix in math. By establishing an equation of the corrected LED array illumination model and the expected irradiance on each point of the lesion, the optimal brightness matrix can be solved according to Eq. (10).
The irradiance of each point on the lesion in this study is expected to be 100 mW∕cm2 but not more than 120 mW∕cm2. Illuminating performance with an all-ones matrix (before optimization) and an optimum brightness matrix (after optimization) are shown in Fig. 4. Figure 4(a) and (b) show different irradiance distributions on a lesion before and after optimization (Pose 1). After imitating the patient’s intra-operation moving from Pose 1, Fig. 4(c) and (d) display another irradiance pattern on the lesion with rotation and translation (Pose 2). The transformation matrix is described as follows:
The effective irradiance proportion (definition: points in the effective irradiance range, 90 mW∕cm2 to 110 mW∕cm2 here, divided by the total points of the lesion) and the coefficient of variation (definition: standard deviation of expected irradiance, 100 mW∕cm2 here, divided by expected irradiance) are two metrics to weigh the effective area and the uniformity of irradiance in PDT. As shown in Fig. 4(e), the statistical result demonstrates that the effective irradiance proportion increases from 84.1% before optimization to 90.9% after optimization for Pose 1, while the coefficient of variation stays unchanged. For Pose 2, the effective irradiance proportion increases significantly by 18.3% (from 66.3% before optimization to 84.6% after optimization) and the coefficient of variation decreases by 1.1%. The results reveal that the corrected LED array illumination model with optimal brightness matrix improves the irradiance uniformity and increases the effective therapy area, even though the patient moves during PDT.
Fig. 4. Comparison of irradiance distribution on the same lesion with different poses before and after optimization. Irradiance distribution on the lesion before (a) and after optimization (b) for Pose 1. Irradiance distribution on the lesion before (c) and after optimization (d) for Pose 2. (e) Statistical assessment of irradiance distribution on the same lesion before and after optimization under two poses.
Due to the dimensions of the lesion exceeding the effective illumination region, a small part of the lesion cannot obtain sufficient light flux. Another regular lesion from the arm is also selected to be conducted on our PDT system, and the lesion area is aimed at the center of the LED array (the pose relationship is shown in Figure S5 in Supplement 1). The lesion is semi-circular in shape with a radius of 8 cm, a length of 8 cm and, a width of 3 cm. Irradiance distributions on the lesion before and after optimization are shown in Fig. 5(a) and (b), respectively. Results demonstrate that the effective irradiance proportion increases dramatically by 24.1% (before optimization: 61.5% vs after optimization: 85.6%), and the coefficient of variation decreases from 9.6% to 6.7%, which reveals that the proposed system is sufficient for improvements of effective irradiance proportion and irradiance uniformity. Figure 5>(c) and (d) display irradiance distributions in detail before and after optimization. After optimization, only 18.7% of the area is exposed to irradiance in excess of 110 mW/cm2 compared to 44.0% of the area before optimization, demonstrating the system's ability to reduce the risk of burns due to high local irradiance.
Fig. 5. Comparison of irradiance distribution on the arm lesion before and after optimization. Irradiance distribution on the lesion before (a) and after optimization (b). Histograms of the irradiance distribution on the lesion before (c) and after optimization (d).
3.5 Light dose monitoring
During PDT by the proposed system, Cam 1 acts as an observer to monitor the patient’s movement. Besides, as the pose relationship between the LED array and Cam 1 has been known, the irradiance distribution on the lesion at any time can be obtained on the premise of establishing an accurate LED array illumination model. Above tasks have been completed in previous sections. By multiplying the irradiance distribution at each pose by the time of treatment, the purpose of monitoring the light flux on the lesion can be achieved, as described in Eq. (12). Figure 6 displays the light flux on the lesion, imitating light flux monitoring in the condition of the patient’s moving during PDT. The patient's lesion is treated in Pose 1 for 5 minutes, light flux is displayed in Fig. 6(a). Then it transforms to Pose 2, after 5 minutes of treatment, the light flux is as shown in Fig. 6(b).
Fig. 6. Light dose monitoring display at different stages. Light dose on the lesion for Pose 1 (a) and Pose 2 (b).
4. Discussions
The dosimetry of PDT is mainly influenced by three factors at the targeted tissue: irradiance and light flux, photosensitizer concentration, and local oxygen concentration [36]. Uniform light flux distribution for patients undergoing photodynamic therapy (PDT) is critical to ensure predictable PDT outcomes [37]. To improve the uniformity of light delivery, many detectors have been developed [38,39] to monitor irradiance. However, since the light flux is a function of treatment time, there is still no effective way to monitor light flux at each point on the lesion currently.
The proposed three-dimensional image-guided photodynamic therapy system with irradiance dynamic planning and flux monitoring may be beneficial for skin disease treatment. To realize monitoring the patient’s movement as widely as possible, the LED array is placed out of the field of view of the 3D camera, and we integrated the LED array coordinate into the 3D camera coordinate system by introducing another 3D camera and performing a registration operation on a reference object. By establishing a corrected LED array illumination model, an optimal brightness matrix can be solved according to the size and shape of the lesion. The proposed system is proven to be able to optimize the irradiance distribution on the lesion surface, even though the lesion moves. Optimization results vary from different poses relationship between the lesion and the LED array. For a proper pose relationship, a traditional LED array could perform well [40], but if patients are moving during PDT, the lesion cannot be fixed in the proper pose [17], the proposed system could ensure the irradiance distribution on the lesion is uniform.
However, an optimal pose relationship is not easy to implement in clinical therapy. For example, part of the lesion does not obtain sufficient light flux, which is partly because of the limited size of the LED array. When treating a lesion with larger area, two strategies are available to perform such treatment. An expanded LED array with a greater field of view 3D camera could be designed to further improve the effective illumination area and monitoring scope. For a larger lesion that may be distributed over the whole body, several proposed systems can be aligned to implement a collaborative therapy with the mosaic technique, where each 3D camera captures its own local image, and these images can be aligned and stitched together for a whole-body image, then the aligned LED arrays can be controlled independently to cover a whole-body treatment.
Irradiance at each point on the lesion can also be calculated due to the corrected LED array illumination model. By integrating with time, the light flux at any point on the lesion under movement can be obtained, which provides an effective method for light dosimetry estimation. Furthermore, as the severity of lesions is also varying in different regions, the demand for light dose may be also different for personalized treatment. As described in Eq. (10), the irradiance of each point on the lesion is expected to be the target irradiance ${E_T}$, we can achieve an optimal solution by updating different expected irradiance value of ${E_T}$ for different severity regions. However, consensus evaluation for the severity of lesions and their dosimetry are the prerequisites. In clinical practice, many physicians are still committing to assessing the severity of lesions to select appropriate treatment methods, performing individualized treatment parameters, and evaluating the treatment effect [41]. For these high-quality clinical trials, our proposed system may provide a new solution for developing a consensus evaluation system on severity evaluation and their optimum irradiance for topical PDT [42].
Furthermore, irradiance-controlling accuracy is limited by the divergence angle of the LED. If a low-cost, narrow-divergence angle LED is developed, controllable irradiance could be theoretically formed on arbitrary surfaces when the LED arrangement is dense enough.
5. Conclusion
This study aims to develop a reliable method for instructing PDT to perform uniform and controllable irradiance on a moving lesion surface. For two typical lesions on the face and arm, the proposed system demonstrates that the effective irradiance proportion and the coefficient of variation increases are both improved, and the irradiance distribution is more concentrated within the safe range. The system can significantly improve the uniformity of irradiance while increasing the effective therapy area and implementing the light flux monitoring. Results shown in our study would provide a new direction to improve procedures controllability of phototherapy for skin diseases.
Funding
National Natural Science Foundation of China (81773349).
Acknowledgements
This research is supported by National Natural Science Foundation of China (Grant No. 81773349).
Disclosures
The authors declare that there are no conflicts of interest related to this article.
Data availability
Code and data generated during this study are available from the corresponding author on reasonable request.
Supplemental document
See Supplement 1 for supporting content.
References
1. S. L. Hagen, K. R. Grey, D. Z. Korta, and K. M. Kelly, “Quality of life in adults with facial port-wine stains,” J. Am. Acad. Dermatol. 76(4), 695–702 (2017). [CrossRef]
2. K. Ezzedine and N. Silverberg, “A practical approach to the diagnosis and treatment of vitiligo in children,” Pediatrics 138(1), e20154126 (2016). [CrossRef]
3. M. Jafferany and P. Pastolero, “Psychiatric and psychological impact of chronic skin disease,” Prim. Care Companion CNS Disord. 20(2), 17nr02247 (2018). [CrossRef]
4. N. W. Nkune and H. Abrahamse, “Nanoparticle-based drug delivery systems for photodynamic therapy of metastatic melanoma: a review,” Int. J. Mol. Sci.22(22), 12549 (2021). [CrossRef]
5. C. Bergqvist and K. Ezzedine, “Vitiligo: a review,” Dermatology 236(6), 571–592 (2020). [CrossRef]
6. J. K. Chen, P. Ghasri, G. Aguilar, A. M. van Drooge, A. Wolkerstorfer, K. M. Kelly, and M. Heger, “An overview of clinical and experimental treatment modalities for port wine stains,” J. Am. Acad. Dermatol. 67(2), 289–304.e29 (2012). [CrossRef]
7. J. K. Nanda, S. W. Dusza, C. Navarrete-Dechent, K. Liopyris, A. A. Marghoob, and M. A. Marchetti, “Incidence of new primary cutaneous melanoma in patients with metastatic melanoma treated with immune checkpoint inhibitors: a single-center cohort study,” JAMA. Dermatol. 157(1), 79–83 (2021). [CrossRef]
8. B. Li, L. Lin, H. Lin, and B. C. Wilson, “Photosensitized singlet oxygen generation and detection: recent advances and future perspectives in cancer photodynamic therapy,” J. Biophotonics 9(11-12), 1314–1325 (2016). [CrossRef]
9. J. Chen, T. Fan, Z. Xie, Q. Zeng, P. Xue, T. Zheng, Y. Chen, X. Luo, and H. Zhang, “Advances in nanomaterials for photodynamic therapy applications: status and challenges,” Biomaterials 237, 119827 (2020). [CrossRef]
10. S. Kwiatkowski, B. Knap, D. Przystupski, J. Saczko, E. Kedzierska, K. Knap-Czop, J. Kotlinska, O. Michel, K. Kotowski, and J. Kulbacka, “Photodynamic therapy - mechanisms, photosensitizers and combinations,” Biomed. Pharmacother. 106, 1098–1107 (2018). [CrossRef]
11. X. Xu, L. Lin, and B. Li, “Automatic protocol for quantifying the vasoconstriction in blood vessel images,” Biomed. Opt. Express 11(4), 2122–2136 (2020). [CrossRef]
12. M. A. MacCormack, “Photodynamic therapy,” Adv. Dermatol. 22, 219–258 (2006). [CrossRef]
13. M. M. Kim and A. Darafsheh, “Light sources and dosimetry techniques for photodynamic therapy,” Photochem. Photobiol. 96(2), 280–294 (2020). [CrossRef]
14. T. Chunhua, G. Liqiang, W. Hua, Z. Jian, N. sili, L. Li, W. Yi, L. Can, L. Xiaoyan, and W. Guanghui, “Efficacy and safety of hemoporfin photodynamic therapy for port-wine stains in paediatric patients: a retrospective study of 439 cases at a single centre,” Photodiagn. Photodyn. Ther. 36, 102568 (2021). [CrossRef]
15. D. Li, X. Nong, Z. Hu, T. Fang, T. Zhao, S. Sun, and Y. Li, “Efficacy and related factors analysis in HMME-PDT in the treatment of port wine stains,” Photodiagnosis. Photodyn. Ther. 29, 101649 (2020). [CrossRef]
16. E. P. M. LaRochelle, M. S. Chapman, E. V. Maytin, T. Hasan, and B. W. Pogue, “Weather-informed light-tissue model-based dose planning for indoor daylight photodynamic therapy,” Photochem. Photobiol. 96(2), 320–326 (2020). [CrossRef]
17. X. M. Hu, F. J. Zhang, F. Dong, and Y. Zhou, “Three-dimensional illumination procedure for photodynamic therapy of dermatology,” J. Biomed. Opt. 19(9), 098003 (2014). [CrossRef]
18. G. Shafirstein, D. A. Bellnier, E. Oakley, S. Hamilton, M. Habitzruther, L. Tworek, A. Hutson, J. A. Spernyak, S. Sexton, L. Curtin, S. G. Turowski, H. Arshad, and B. Henderson, “Irradiance controls photodynamic efficacy and tissue heating in experimental tumours: implication for interstitial PDT of locally advanced cancer,” Br. J. Cancer 119(10), 1191–1199 (2018). [CrossRef]
19. Y. Hu, K. Wang, and T. C. Zhu, “A light blanket for intraoperative photodynamic therapy,” in Proc SPIE Int. Soc. Opt. Eng., (SPIE, 2009), 73801W.
20. B. Selm, M. Rothmaier, M. Camenzind, T. Khan, and H. Walt, “Novel flexible light diffuser and irradiation properties for photodynamic therapy,” J. Biomed. Opt. 12(3), 034024 (2007). [CrossRef]
21. Y. Oguz, V. Koncar, C. Cochrane, and S. Mordon, “Light-emitting woven fabric for treatment with photodynamic therapy and monitoring of actinic keratosis,” in Photomedicine - Advances in Clinical Practice, Y. Tanaka, ed. (IntechOpen, London, United Kingdom, 2017), pp. 95–111.
22. B. W. Henderson, T. M. Busch, L. A. Vaughan, N. P. Frawley, D. Babich, T. A. Sosa, J. D. Zollo, A. S. Dee, M. T. Cooper, D. A. Bellnier, W. R. Greco, and A. R. Oseroff, “Photofrin photodynamic therapy can significantly deplete or preserve oxygenation in human basal cell carcinomas during treatment, depending on fluence rate,” Cancer. Res. 60(3), 525–529 (2000).
23. D. Chen, H. Zheng, Z. Huang, H. Lin, Z. Ke, S. Xie, and B. Li, “Light-emitting diode-based illumination system for in vitro photodynamic therapy,” Int. J. Photoenergy 2012, 1–6 (2012). [CrossRef]
24. A. K. L. Fujita, D. J. Chianfrome, V. S. Moreira, A. L. Zanchin, P. F. C. D. Menezes, and V. S. Bagnato, “Anatomically adjustable device for large-area photodynamic therapy,” in Photodynamic Therapy - From Basic Science to Clinical Research, N. Inada, ed. (IntechOpen, London, United Kingdom, 2021), pp. 1–13.
25. M. Panjehpour, R. C. DeNovo, M. G. Petersen, B. F. Overholt, R. Bower, V. Rubinchik, and B. Kelly, “Photodynamic therapy using verteporfin (benzoporphyrin derivative monoacid ring A, BPD-MA) and 630 nm laser light in canine esophagus,” Laser. Surg. Med. 30(1), 26–30 (2002). [CrossRef]
26. J. M. Ang, I. B. Riaz, M. U. Kamal, G. Paragh, and N. C. Zeitouni, “Photodynamic therapy and pain: a systematic review,” Photodiagnosis. Photodyn. Ther. 19, 308–344 (2017). [CrossRef]
27. Y. Hongming, J. W. M. Bergmans, T. C. W. Schenk, J. P. M. G. Linnartz, and R. Rietman, “Uniform illumination rendering using an array of LEDs: a signal processing perspective,” IEEE Trans. Signal Process. 57(3), 1044–1057 (2009). [CrossRef]
28. G. Praneeth Varma, R. Sushma, V. Sharma, A. Kumar, and G. V. V. Sharma, “Power allocation for uniform illumination with stochastic LED arrays,” Opt. Express 25(8), 8659–8669 (2017). [CrossRef]
29. M. Le. Tran and S. Kim, “Joint power allocation and orientation for uniform illuminance in indoor visible light communication,” Opt. Express 27(20), 28575–28587 (2019). [CrossRef]
30. Z. Su, D. Xue, and Z. Ji, “Designing LED array for uniform illumination distribution by simulated annealing algorithm,” Opt. Express 20(S6), A843–855 (2012). [CrossRef]
31. Z. Qin, K. Wang, F. Chen, X. Luo, and S. Liu, “Analysis of condition for uniform lighting generated by array of light emitting diodes with large view angle,” Opt. Express 18(16), 17460–17476 (2010). [CrossRef]
32. K. Pulli, “Multiview registration for large data sets,” in Second international conference on 3-d digital imaging and modeling, (IEEE, Ottawa, Canada, 1999).
33. I. Moreno, M. Avendaño-Alejo, and R. I. Tzonchev, “Designing light-emitting diode arrays for uniform near-field irradiance,” Appl. Opt. 45(10), 2265–2272 (2006). [CrossRef]
34. X. Wang, J. Xiong, X. Hu, and Q. Li, “Implementation and uniformity calibration of LED array for photodynamic therapy,” J. Innov. Opt. Heal. Sci. 15(04), 2240004 (2022). [CrossRef]
35. W. Kang, X. Wang, J. Zhang, X. Hu, and Q. Li, “Two-way perceived color difference saliency algorithm for image segmentation of port wine stains,” in Intelligent Computing and Block Chain1385 (Springer, 2021), pp. 50–60.
36. H. Liu, “Editorial: dosimetry study in photodynamic therapy for diagnosis, precision treatment and treatment evaluation,” Front. Phys. 8, 588484 (2020). [CrossRef]
37. T. C. Zhu, Y. Ong, M. M. Kim, X. Liang, J. C. Finlay, A. Dimofte, C. B. Simone 2nd, J. S. Friedberg, T. M. Busch, E. Glatstein, and K. A. Cengel, “Evaluation of light fluence distribution using an IR navigation system for HPPH-mediated pleural photodynamic therapy (pPDT),” Photochem. Photobiol. 96(2), 310–319 (2020). [CrossRef]
38. M. M. Kim, T. C. Zhu, Y. H. Ong, J. C. Finlay, A. Dimofte, S. Singhal, E. Glatstein, and K. A. Cengel, “Infrared navigation system for light dosimetry during pleural photodynamic therapy,” Phys. Med. Biol. 65(7), 075006 (2020). [CrossRef]
39. L. Lilge, J. Wu, Y. Xu, A. Manalac, D. Molenhuis, F. Schwiegelshohn, L. Vesselov, W. Embree, M. Nesbit, V. Betz, A. Mandel, M. A. S. Jewett, and G. S. Kulkarni, “Minimal required PDT light dosimetry for nonmuscle invasive bladder cancer,” J. Biomed. Opt. 25(06), 1 (2020). [CrossRef]
40. F. Zhang, X. Hu, Y. Zhou, and Q. Li, “Optimization of irradiance for photodynamic therapy of port-wine stain,” J. Biomed. Opt. 20(4), 048004 (2015). [CrossRef]
41. J. Ren, H. Tuan, C. Huang, D. Shu, D. Chen, E. Y. Zhou, D. Liu, P. Tu, and Y. Zhao, “A proposed scoring system for facial port-wine stain evaluation: facial port-wine stain area and severity index,” J. Cosmet. Dermatol. 21(7), 2931–2938 (2022). [CrossRef]
42. M. I. van Raath, S. Chohan, A. Wolkerstorfer, C. van der Horst, J. Limpens, X. Huang, B. Ding, G. Storm, R. van der Hulst, and M. Heger, “Clinical outcome measures and scoring systems used in prospective studies of port wine stains: a systematic review,” PLoS One 15(7), e0235657 (2020). [CrossRef]
