Tesis

Penggunaan Kecerdasan Buatan untuk Penilaian ASPECT Score pada Stroke Iskemik Akut: Telaah Sistematis dan Meta Analisis = The Use of Artificial Intelligence for ASPECT Score Assesment in Acute Ischemic Stroke: A Systematic Review and Meta Analysis.

Pendahuluan: Stroke merupakan suatu defisit neurologis fokal yang bersifat akut atau terjadi secara mendadak disebabkan oleh lesi pada vaskular. Onset terjadinya stroke mendadak dan memiliki simptom yang dapat bertahan hingga lebih dari 24 jam. Alberta Stroke Program Early Computed Tomography Score (ASPECTS) adalah sistem perhitungan semikuantitatif yang berperan dalam menilai luasnya gambaran stroke iskemik, dengan skor ≥ 6 yang merupakan cutoff point penting. Metode: Telaah sistematis dilakukan dengan mengacu pada panduan Prefered Reporting Items for Systematic Review and Meta-analyses (PRISMA) 2020. Pencarian utama dilakukan melalui pangkalan data dengan menggunakan beberapa kata kunci yang telah ditentukan dalam kurun waktu kurang dari 10 tahun terakhir terhitung sejak 2024 pada database PubMed, Cochrane dan ProQuest. Peserta studi merupakan pasien-pasien dengan suspek stroke iskemik akut yang dilakukan pemeriksaan CT Scan cerebral non kontras dalam penilaian ASPECT Score oleh Dokter spesialis Radiologi dan subspesialis Neuroradiologi menggunakan software Brainomax e-ASPECTS, iSchemaView RAPID ASPECTS, dan Syngo via Frontier ASPECTS). Hasil: Secara garis besar model AI yang dikembangkan oleh Kuang et al. (2020) cukup baik dalam mengidentifikasi kasus positif dan negatif dengan tepat, karena menghasilkan nilai sensitivitas dan spesifisitas yang relatif balanced, yaitu nilai sensitivitas sebesar 0,66 (CI 95%: 0,60–0,72) dan spesifisitas sebesar 0,92 (CI 95%: 0,90–0,94). Selain itu, model AI yang dikembangkan oleh Joon Lee et al. (2022) juga bisa jadi alternatif lain dalam mengidentifikasi kasus positif dan negatif dengan tepat, karena menghasilkan nilai sensitivitas dan spesifisitas yang juga relatif balanced, yaitu nilai sensitivitas sebesar 0,63 (CI 95%: 0,58–0,67) dan spesifisitas sebesar 0,97 (CI 95%: 0,96–0,97). Kesimpulan: Kecerdasan buatan berpotensi meningkatkan akurasi diagnostik ASPECTS sehingga diharapkan dapat membantu Dokter spesialis Radiologi dalam meningkatkan efisiensi waktu pembacaan final CT scan kasus stroke iskemik akut,.
Kata kunci: ASPECTS, Artificial Intelligence in Radiology, Acute Ischemic Stroke, dan Neuroadiology


Introduction: Stroke is a focal neurological deficit that is acute or occurs suddenly due to vascular lesions. The onset of stroke is sudden and has symptoms that can last for more than 24 hours. The Alberta Stroke Program Early Computed Tomography Score (ASPECTS) is a semi-quantitative calculation system that plays a role in assessing the extent of ischemic stroke images, with a score ≥ 6 which is an important cutoff point. Methods: A systematic review was conducted with reference to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) 2020 guidelines. The index test in this systematic review was ASPECTS. The target condition was patients with neurological deficits suspected of acute ischemic stroke. The standard reference used was a non-contrast cerebral CT (NCCT). The primary search was conducted through databases using keywords that had been determined in a period of less than 10 years since 2024 in the scientific databases PubMed, Cochrane and ProQuest. The study participants included were patients with suspected acute ischemic stroke who underwent NCCT in the ASPECT Score assessment by Radiology specialists and Neuroradiology subspecialists and using software Brainomax e-ASPECTS, iSchemaView RAPID ASPECTS, and Syngo via Frontier ASPECTS). Results: In general, the AI model developed by Kuang et al. (2020) is quite good at identifying positive and negative cases accurately, because it produces relatively balanced sensitivity and specificity values, namely a sensitivity value of 0.66 (CI 95%: 0.60–0.72) and a specificity of 0.92 (CI 95%: 0.90–0.94). In addition, the AI model developed by Joon Lee et al. (2022) can also be another alternative in identifying positive and negative cases accurately, because it produces sensitivity and specificity values that are also relatively balanced, namely a sensitivity value of 0.63 (CI 95%: 0.58–0.67) and a specificity value of 0.97 (CI 95%: 0.96–0.97). Conclusion: Artificial intelligence has great potential in improving the diagnostic accuracy of ASPECTS so that it is expected to help Radiology specialists in increasing the efficiency of the reading process time of the final CT scan assessment in acute ischemic stroke cases.
Keywords: ASPECTS, Artificial Intelligence in Radiology, Acute Ischemic Stroke, and Neuroradiology.