RESUMO
We developed a surgical support system that visualises important microanatomies using artificial intelligence (AI). This study evaluated its accuracy in recognising the thoracic nerves during lung cancer surgery. Recognition models were created with deep learning using images precisely annotated for nerves. Computational evaluation was performed using the Dice index and the Jaccard index. Four general thoracic surgeons evaluated the accuracy of nerve recognition. Further, the differences in time lag, image quality and smoothness of movement between the AI system and surgical monitor were assessed. Ratings were made using a five-point scale. The computational evaluation was relatively favourable, with a Dice index of 0.56 and a Jaccard index of 0.39. The AI system was used for 10 thoracoscopic surgeries for lung cancer. The accuracy of thoracic nerve recognition was satisfactory, with a recall score of 4.5 ± 0.4 and a precision score of 4.0 ± 0.9. Though smoothness of motion (3.2 ± 0.4) differed slightly, nearly no difference in time lag (4.9 ± 0.3) and image quality (4.6 ± 0.5) between the AI system and the surgical monitor were observed. In conclusion, the AI surgical support system has a satisfactory accuracy in recognising the thoracic nerves.
Assuntos
Inteligência Artificial , Nervos Torácicos , Humanos , Neoplasias Pulmonares/cirurgia , Aprendizado Profundo , Cirurgia Assistida por Computador/métodos , Processamento de Imagem Assistida por Computador/métodosRESUMO
The advent of Artificial Intelligence (AI)-based object detection technology has made identification of position coordinates of surgical instruments from videos possible. This study aimed to find kinematic differences by surgical skill level. An AI algorithm was developed to identify X and Y coordinates of surgical instrument tips accurately from video. Kinematic analysis including fluctuation analysis was performed on 18 laparoscopic distal gastrectomy videos from three expert and three novice surgeons (3 videos/surgeon, 11.6 h, 1,254,010 frames). Analysis showed the expert surgeon cohort moved more efficiently and regularly, with significantly less operation time and total travel distance. Instrument tip movement did not differ in velocity, acceleration, or jerk between skill levels. The evaluation index of fluctuation ß was significantly higher in experts. ROC curve cutoff value at 1.4 determined sensitivity and specificity of 77.8% for experts and novices. Despite the small sample, this study suggests AI-based object detection with fluctuation analysis is promising because skill evaluation can be calculated in real time with potential for peri-operational evaluation.
Assuntos
Inteligência Artificial , Competência Clínica , Gastrectomia , Laparoscopia , Laparoscopia/métodos , Humanos , Gastrectomia/métodos , Gravação em Vídeo/métodos , Masculino , Feminino , Algoritmos , Fenômenos Biomecânicos , Curva ROCRESUMO
BACKGROUND: A postoperative pancreatic fistula (POPF) is a critical complication of radical gastrectomy for gastric cancer, mainly because surgeons occasionally misrecognize the pancreas and fat during lymphadenectomy. Therefore, this study aimed to develop an artificial intelligence (AI) system capable of identifying and highlighting the pancreas during robot-assisted gastrectomy. METHODS: A pancreas recognition algorithm was developed using HRNet, with 926 training images and 232 validation images extracted from 62 scenes of robot-assisted gastrectomy videos. During quantitative evaluation, the precision, recall, intersection over union (IoU), and Dice coefficients were calculated based on the surgeons' ground truth and the AI-inferred image from 80 test images. During the qualitative evaluation, 10 surgeons answered two questions related to sensitivity and similarity for assessing clinical usefulness. RESULTS: The precision, recall, IoU, and Dice coefficients were 0.70, 0.59, 0.46, and 0.61, respectively. Regarding sensitivity, the average score for pancreas recognition by AI was 4.18 out of 5 points (1 = lowest recognition [less than 50%]; 5 = highest recognition [more than 90%]). Regarding similarity, only 54% of the AI-inferred images were correctly differentiated from the ground truth. CONCLUSIONS: Our surgical AI system precisely highlighted the pancreas during robot-assisted gastrectomy at a level that was convincing to surgeons. This technology may prevent misrecognition of the pancreas by surgeons, thus leading to fewer POPFs.
Assuntos
Inteligência Artificial , Gastrectomia , Pâncreas , Procedimentos Cirúrgicos Robóticos , Neoplasias Gástricas , Humanos , Gastrectomia/métodos , Procedimentos Cirúrgicos Robóticos/métodos , Neoplasias Gástricas/cirurgia , Pâncreas/cirurgia , Algoritmos , Fístula Pancreática/etiologia , Complicações Pós-Operatórias , CirurgiõesRESUMO
This retrospective study assessed the treatment planning data and clinical outcomes for 152 prostate cancer patients: 76 consecutive patients treated by carbon-ion radiation therapy and 76 consequtive patients treated by moderate hypo-fractionated intensity-modulated photon radiation therapy. These two modalities were compared using linear quadratic model equivalent doses in 2 Gy per fraction for rectal or rectal wall dose-volume histogram, 3.6 Gy per fraction-converted rectal dose-volume histogram, normal tissue complication probability model, and actual clinical outcomes. Carbon-ion radiation therapy was predicted to have a lower probability of rectal adverse events than intensity-modulated photon radiation therapy based on dose-volume histograms and normal tissue complication probability model. There was no difference in the clinical outcome of rectal adverse events between the two modalities compared in this study.
Assuntos
Neoplasias da Próstata , Radioterapia Conformacional , Radioterapia de Intensidade Modulada , Carbono , Humanos , Masculino , Probabilidade , Neoplasias da Próstata/radioterapia , Dosagem Radioterapêutica , Planejamento da Radioterapia Assistida por Computador , Radioterapia de Intensidade Modulada/efeitos adversos , Reto , Estudos RetrospectivosRESUMO
We investigated the normal tissue complication probability (NTCP) of the incidence of pericardial effusion (PCE) based on the mean heart dose (MHD) in patients with oesophageal cancer treated with definitive chemoradiotherapy. The incidences of PCE in any grade (A-PCE) and symptomatic PCE (S-PCE) were evaluated separately. To identify predictors for PCE, several clinical and dose-volume parameters were analysed using a receiver operating characteristic (ROC) curve and multivariate regression analysis. To validate its clinical applicability, the generated NTCP model was compared to the Lyman-Kutcher-Burman (LKB) model. Among 229 eligible patients, A-PCE and S-PCE were observed in 100 (43.7%) and 18 (7.9%) patients, respectively. MHD showed a preferable area under the curve (AUC) value for S-PCE (AUC = 0.821) and A-PCE (AUC = 0.734). MHD was the only significant predictor for A-PCE; MHD and hypertension were selected as significant factors for S-PCE. The estimated NTCP, using the MHD-based model, showed excellent correspondence to the LKB model in A-PCE and S-PCE. The NTCP curve of A-PCE was gentler than that of S-PCE and had no threshold. The MHD-based NTCP model was simple but comparable to the LKB model for both A-PCE and S-PCE. Therefore, the estimated NTCP may provide clinically useful parameters for predicting PCE.
Assuntos
Relação Dose-Resposta à Radiação , Neoplasias Esofágicas/complicações , Coração/efeitos da radiação , Modelos Biológicos , Derrame Pericárdico/complicações , Probabilidade , Adulto , Idoso , Idoso de 80 Anos ou mais , Feminino , Humanos , Modelos Logísticos , Masculino , Pessoa de Meia-IdadeRESUMO
BACKGROUND AND PURPOSE: We aimed to evaluate the relationship between brainstem necrosis and dose-volume histograms in patients with head and neck tumors after carbon-ion radiotherapy. MATERIAL AND METHODS: We evaluated 85 patients with head and neck tumors who underwent carbon-ion radiotherapy and were followed-up for ≥12months. Brainstem necrosis was evaluated using the Common Terminology Criteria for Adverse Events (version 4.0). RESULTS: The median follow-up was 24months, and four patients developed grade 1 brainstem necrosis, with 2-year and 3-year cumulative rates of 2.8% and 6.5%, respectively. Receiver operating characteristic curve analysis revealed the following significant cut-off values: a maximum brainstem dose of 48Gy (relative biological effectiveness [RBE]), D1cm3 of 27Gy (RBE), V40Gy (RBE) of 0.1cm3, V30Gy (RBE) of 0.7cm3, and V20Gy (RBE) of 1.4cm3. Multivariate analysis revealed that V30Gy (RBE) was most significantly associated with brainstem necrosis. The 2-year cumulative rates were 33% and 0% for V30Gy (RBE) of ≥0.7cm3 and <0.7cm3, respectively (p<0.001). CONCLUSIONS: The present study indicated that the dose constraints might help minimize brainstem necrosis after carbon-ion radiotherapy.
Assuntos
Tronco Encefálico/patologia , Neoplasias de Cabeça e Pescoço/radioterapia , Radioterapia com Íons Pesados , Adulto , Idoso , Idoso de 80 Anos ou mais , Feminino , Neoplasias de Cabeça e Pescoço/patologia , Radioterapia com Íons Pesados/efeitos adversos , Humanos , Masculino , Pessoa de Meia-Idade , Necrose , Curva ROC , Dosagem Radioterapêutica , Eficiência Biológica RelativaRESUMO
The safety and efficacy of carbon-ion radiotherapy for advanced non-small cell lung cancer have not been established. We evaluated the clinical outcomes and dose-volume histogram parameters of carbon-ion radiotherapy compared with photon therapy in T2b-4N0M0 non-small cell lung cancer. Twenty-three patients were treated with carbon-ion radiotherapy between May 2011 and December 2015. Seven, 14, and 2 patients had T2b, T3, and T4, respectively. The median age was 78 (range, 53-91) years, with 22 male patients. There were 12 adenocarcinomas, 8 squamous cell carcinomas, 1 non-small cell lung carcinoma, and 2 clinically diagnosed lung cancers. Eleven patients were operable, and 12 patients were inoperable. Most patients (91%) were treated with carbon-ion radiotherapy of 60.0 Gy relative biological effectiveness (RBE) in 4 fractions or 64.0 Gy (RBE) in 16 fractions. Local control and overall survival rates were calculated. Dose-volume histogram parameters of normal lung and tumor coverages were compared between carbon-ion radiotherapy and photon therapies, including three-dimensional conformal radiotherapy (3DCRT) and intensity-modulated radiotherapy (IMRT). The median follow-up of surviving patients was 25 months. Three patients experienced local recurrence, and the 2-year local control rate was 81%. During follow-up, 5 patients died of lung cancer, and 1 died of intercurrent disease. The 2-year overall survival rate was 70%. Operable patients had a better overall survival rate compared with inoperable patients (100% vs. 43%; P = 0.04). There was no grade ≥2 radiation pneumonitis. In dose-volume histogram analysis, carbon-ion radiotherapy had a significantly lower dose to normal lung and greater tumor coverage compared with photon therapies. Carbon-ion radiotherapy was effectively and safely performed for T2b-4N0M0 non-small cell lung cancer, and the dose distribution was superior compared with those for photon therapies. A Japanese multi-institutional study is ongoing to prospectively evaluate these patients and establish the use of carbon-ion radiotherapy.
Assuntos
Carcinoma Pulmonar de Células não Pequenas/radioterapia , Radioterapia com Íons Pesados , Neoplasias Pulmonares/radioterapia , Fótons , Idoso , Idoso de 80 Anos ou mais , Feminino , Radioterapia com Íons Pesados/efeitos adversos , Humanos , Masculino , Pessoa de Meia-Idade , Projetos Piloto , Taxa de Sobrevida , Resultado do TratamentoRESUMO
Radiotherapy using charged particles is characterized by a low dose to the surrounding healthy organs, while delivering a high dose to the tumor. However, interfractional anatomical changes can greatly affect the robustness of particle therapy. Therefore, we compared the dosimetric impact of interfractional anatomical changes (i.e. body contour differences and gastrointestinal gas volume changes) in photon, proton and carbon ion therapy for pancreatic cancer patients. In this retrospective planning study, photon, proton and carbon ion treatment plans were created for 9 patients. Fraction dose calculations were performed using daily cone-beam CT (CBCT) images. To this end, the planning CT was deformably registered to each CBCT; gastrointestinal gas volumes were delineated on the CBCTs and copied to the deformed CT. Fraction doses were accumulated rigidly. To compare planned and accumulated dose, dose-volume histogram (DVH) parameters of the planned and accumulated dose of the different radiotherapy modalities were determined for the internal gross tumor volume, internal clinical target volume (iCTV) and organs-at-risk (OARs; duodenum, stomach, kidneys, liver and spinal cord). Photon plans were highly robust against interfractional anatomical changes. The difference between the planned and accumulated DVH parameters for the photon plans was less than 0.5% for the target and OARs. In both proton and carbon ion therapy, however, coverage of the iCTV was considerably reduced for the accumulated dose compared with the planned dose. The near-minimum dose ([Formula: see text]) of the iCTV reduced with 8% for proton therapy and with 10% for carbon ion therapy. The DVH parameters of the OARs differed less than 3% for both particle modalities. Fractionated radiotherapy using photons is highly robust against interfractional anatomical changes. In proton and carbon ion therapy, such changes can severely reduce the dose coverage of the target.
Assuntos
Radioterapia com Íons Pesados/métodos , Neoplasias Pancreáticas/radioterapia , Terapia com Prótons/métodos , Doses de Radiação , Radioterapia de Intensidade Modulada/métodos , Radioisótopos de Carbono/uso terapêutico , Humanos , Órgãos em Risco , Fótons/uso terapêutico , Dosagem RadioterapêuticaRESUMO
A three-dimensional dosimetry method is strongly required in the dose distribution measurement of a patient QA of a heavy ion therapy. Nanocomposite Fricke gel dosimeters are the most possible candidate for this purpose. Experimental dose distribution measurements were carried out using a scanning irradiation port of Gunma University Heavy Ion Medical Center. The result showed no significant LET dependence and indicated a possibility for a precise dosimetry of a heavy ion therapy. It also indicated the importance of three-dimensional dosimetry in the commissioning process of the treatment accelerator.
Assuntos
Carbono , Dosímetros de Radiação , Humanos , RadiometriaRESUMO
BACKGROUND AND PURPOSE: We evaluated the robustness of carbon ion therapy for pancreatic cancer patients by investigating the impact of interfractional anatomical changes on the accumulated dose when using bony anatomy- and fiducial marker-based position verification. MATERIAL AND METHODS: Carbon ion treatment plans were created for 9 patients in this retrospective planning study. The planning CT was deformably registered to each daily cone-beam CT (CBCT). The gastrointestinal gas volume visible on each CBCT was copied to these deformed CT images. Subsequently, the fraction doses were calculated by aligning the treatment plan according to a bony anatomy- and a fiducial marker-based registration. We compared the accumulated fraction doses with the planned dose using dose-volume histograms (DVHs) of the internal gross tumour volume (iGTV), internal clinical target volume (iCTV), duodenum, stomach, liver, spinal cord and kidneys. RESULTS: iCTV coverage (D98%) was on average reduced from 98.6% as planned to 81.9% and 88.6% for the bony anatomy- and marker-based registrations, respectively. DVHs of the duodenum showed large differences between the planned and accumulated dose. CONCLUSIONS: Severe reductions in dose coverage of the target due to interfractional anatomical changes were observed in both position verification methods.