Àá½Ã¸¸ ±â´Ù·Á ÁÖ¼¼¿ä. ·ÎµùÁßÀÔ´Ï´Ù.

Æó¾Ï°ËÁø¿¡¼­ ÀΰøÁö´É ±â¼úÀÇ È°¿ë Application of Artificial Intelligence in Lung Cancer Screening

´ëÇÑ¿µ»óÀÇÇÐȸÁö 2019³â 80±Ç 5È£ p.872 ~ 879
ÀÌ»ó¹Î, ¹Úâ¹Î,
¼Ò¼Ó »ó¼¼Á¤º¸
ÀÌ»ó¹Î ( Lee Sang-Min ) 
¿ï»ê´ëÇб³ ÀÇ°ú´ëÇÐ ¼­¿ï¾Æ»êº´¿ø ¿µ»óÀÇÇаú

¹Úâ¹Î ( Park Chang-Min ) 
¼­¿ï´ëÇб³ ÀÇ°ú´ëÇÐ ÀÇÇבּ¸¿ø ¹æ»ç¼±ÀÇÇבּ¸¼Ò

Abstract

Àú¼±·® CT¸¦ ÀÌ¿ëÇÑ Æó¾Ï°ËÁøÀº Æó¾Ï »ç¸Á·ü °¨¼Ò È¿°ú°¡ ÀÔÁõµÇ¾úÀ¸¸ç, ±¹³»¿¡¼­µµ ½ÃÀ۵Ǿú´Ù. È¿°úÀûÀÎ Æó¾Ï°ËÁøÀ» À§Çؼ­´Â Àú¼±·® CT¿¡ ´ëÇÑ Á¤È®ÇÑ Æǵ¶ÀÌ ÀüÁ¦µÇ¾î¾ß ÇÑ´Ù. ÇÏÁö¸¸, ÃßÁ¤ °Ë»ç °Ç¼ö¿Í °æÇè ÀÖ´Â Àü¹®°¡ÀÇ ¼ö¸¦ °í·ÁÇßÀ» ¶§ ¿µ»óÀÇÇÐ Áø·á¿¡ Å« ºÎ´ãÀÌ µÉ °ÍÀº ÀÚ¸íÇØ º¸ÀδÙ. ÀÌ ¹®Á¦ÀÇ ÇØ°á Ãø¸é¿¡¼­, ÀΰøÁö´É ±â¼úÀ» È°¿ëÇÑ Àú¼±·® CT Æǵ¶ º¸Á¶½Ã½ºÅÛ °³¹ß°ú Àû¿ë¿¡ Çаè¿Í °ü·Ã »ê¾÷°èÀÇ °ü½ÉÀÌ ¸ð¾ÆÁö°í ÀÖ´Ù. ƯÈ÷, ÀÇÇпµ»ó ºÐ¼®¿¡ Á÷Á¢ Àû¿ëÀÌ °¡´ÉÇÑ µö·¯´×(deep learning) ±â¼úÀº ±âÁ¸ ±â°èÇнÀ(machine learning) ±â¼úº¸´Ù ¿ì¼öÇÑ Áø´Ü ¼º´ÉÀ» º¸ÀÌ°í ÀÖ¾î, ±× ÀáÀçÀû À¯¿ë¼º¿¡ ´ëÇÑ ¿¬±¸°¡ È°¹ßÈ÷ ÁøÇàµÇ°í ÀÖ´Ù. Æó¾Ï°ËÁø¿¡¼­ µö·¯´×À» Æ÷ÇÔÇÑ ÀΰøÁö´É ±â¼úÀ» Àû¿ëÇÒ ¼ö ÀÖ´Â ºÐ¾ß´Â Å©°Ô ÄÄÇ»ÅÍ º¸Á¶ º´º¯ °ËÃâ, Æǵ¶¹® »ý¼º, °ËÃâµÈ Æó°áÀýÀÇ ¾Ç¼ºµµ Æò°¡, ±×¸®°í ȯÀÚÀÇ ¿¹ÈÄ ¿¹ÃøÀ¸·Î ³ª´©¾î º¼ ¼ö ÀÖ´Ù. ÀÌ¿¡ º» ±â°í¹®¿¡¼­´Â ÇöÀç Æó¾Ï°ËÁø¿¡ È°¿ëÇÒ ¼ö ÀÖ´Â ÀΰøÁö´É ±â¹Ý ¿¬±¸µéÀ» »ìÆ캸°í, ÇâÈÄ À̸¦ ÀÌ¿ëÇÑ Æó¾Ï°ËÁøÀÇ °¡´É¼º¿¡ ´ëÇØ ³íÀÇÇÏ°íÀÚ ÇÑ´Ù.

Lung cancer is a leading cause of deaths due to cancer, worldwide. At present, low-dose computed tomography (CT) is the only established screening method for reducing lung cancer mortality. However, several challenges must be overcome, to ensure the implementation of lung cancer screening, which include a large number of expected low-dose CT examinations and relative shortage of experienced radiologists for interpreting them. The use of artificial intelligence has garnered attention in this regard. A deep learning technique, which is a subclass of machine learning methods, involving the learning of data representations in an end-to-end manner, has already demonstrated outstanding performance in medical image analysis. Several studies are exploring the possibility of deep learning-based applications in medical domains, including radiology. In lung cancer screening, computer-aided detection, report generation, prediction of malignancy in the detected nodules, and prognosis prediction can be considered for the application of artificial intelligence. This article will cover the current status of deep learning approaches, their limitations, and their potential in lung cancer screening programs.

Å°¿öµå

Lung Neoplasms; Screening; Computed Tomography, X-Ray; Artificial Intelligence

¿ø¹® ¹× ¸µÅ©¾Æ¿ô Á¤º¸

 

µîÀçÀú³Î Á¤º¸

KCI
KoreaMed
KAMS