Stainless steel is widely used in bone fixation, cardiovascular systems, catheters, surgical instruments and dental crowns (30). Furthermore, Bhushan et al (27) studied the effect of stainless-steel hip prosthesis on radiation using a customized prosthesis containing wrought austenitic stainless steel. It was observed that for 6 MV of photon irradiation, at a depth of 10 cm below the prosthesis, with field sizes of 5x5, 10x10 and 20x20, the dose attenuation was 8.3, 7.4 and 7.5% when the prosthesis was present compared with in its absence. In addition, when the energy was increased to 15 MV, the dose attenuations were 7.6, 7.1 and 5.0% for the same distances 5x5, 10x10 and 20x20, respectively, of the field sizes. Moreover, Mahuvava and Du Plessis (25) observed that when bilateral stainless steel hips were present, the attenuation of radiation by a prosthesis was 22.8, 20.4, 18.5 and 16.9% with photon irradiations at 6, 10, 15 and 20 MV, respectively. Furthermore, Liu et al (23) used human cadavers to simulate tumor resection for internal fixation surgery by placing stainless steel plates in the anterior and upper 1/3 of the human femur with a muscle strip of the same size and thickness for control purposes. It was observed that the absorbed dose at the incident surface increased by 21.65%. Conversely, the absorbed dose at the exit surface was attenuated by 8.42% compared with the control group, as measured with a pyroelectric dosimeter under 6 MV X-ray irradiation. Additionally, their experiments also used the treatment planning system (TPS). It was observed that the distance from the tissue to the metal surface was an important factor affecting dose absorption, and this effect was greatest at a distance of 0.5 cm from the metal surface, resulting in a 6.1% increase in dose upstream and a 2.2% dose attenuation downstream. In addition, He and Ni (26) used the Monte Carlo (MC) algorithm to simulate 6 MV X-ray irradiation and observed that the incident surface dose of stainless steel implants with thicknesses of 1, 2 and 4 cm increased by 23.8, 24.0 and 24.3%, respectively, compared with the dose without the implant; by contrast, the dose at the exit surface decreased by 23.0, 35.2 and 55.1%, respectively. This indicates that the thickness of the implant did not significantly affect the incident radiation dose; however, the dose attenuated more with increasing metal thickness at the exit surface.

Stainless steel is majorly used in stents. Chen et al (5) used a solid water phantom to simulate the tissue environment of the human esophagus and measured the surrounding irradiation dose using thermo-luminescent dosimeters. It was observed that the increase in dose to the Z-stent's (stainless steel) anterior surface was 3.5-7.8% when using single beams. Abu Dayyeh et al (19) used a solid water phantom irradiated with 6, 10 and 18 MV photons and observed that the dose enhancements of the stainless-steel stent Wallflex front upstream were 4.2, 5.2 and 6.7%, respectively, at three different photon energy irradiations. The aforementioned experiments revealed that the presence of stainless-steel implants could have a non-negligible effect on the accuracy of the RT dose; therefore, when stainless steel implants are present in the irradiation field, this dose perturbation should be considered when making RT plans.

Since the introduction of pure titanium for oral implants in the 1960s, it has been widely used as a material for surgical implants. Subsequently, Ti-3Al-2.5V and Ti-6Al-4V were gradually used as femoral and tibial replacement materials. Ti₆Al₄V is an important titanium alloy widely used as a material in surgical implants (31).

External irradiation is a common treatment modality for prostate cancer. With an aging population, several patients treated for prostate or pelvic tumors undergo partial or total hip replacement because of osteoarthritis and hip dysfunction (27). Titanium is often used for hip implants because of its excellent biomedical properties, however it inevitably may affect dose delivery (32).

Ade and du Plessis (29) investigated the perturbation effect of a unilateral titanium prosthesis on the dose distribution of 6 MV and 15 MV photon beams. Using a built-in titanium hip prosthesis, it was observed that the proximal dose enhancement of the prosthesis ranged as 21-23%, and the distal dose reduction of the prosthesis was 18-21% with 6 MV photon beam irradiation. However, when the radiation energy was increased to 25 MV, the proximal dose enhancement and distal dose attenuation were 25-30 and 15-18%, respectively. It could be inferred from their experiments that the field size does not significantly affect the dose, and that the most significant dose change is within 1.0 cm from the implant surface. Additionally, Akyol et al (8) used the pencil beam convolution (PBC) algorithm of the TPS and MC simulation techniques for their study. They simulated a linear accelerator to produce a 6-MV photon beam and observed an 11.2% increase in dose anterior to the titanium dental implant and a 15.5% decrease in dose posteriorly.

Furthermore, He and Ni (26) used an MC algorithm to simulate 6-MV X-ray irradiation and compared the effect of different thicknesses of stainless-steel plates with that of titanium metal implants on the radiation dose. It was observed that titanium implants increased the upstream dose by 19.8, 20.3 and 20.6% at the thicknesses of 1, 2 and 4 cm, respectively, while decreasing the downstream dose by 18.4, 23.6 and 35.0%, respectively. Additionally, it was observed that the effect of titanium implants on radiation dose was less than that of stainless-steel implants. Similarly, experiments by Liu et al (23) on human cadavers revealed that the effect of stainless-steel plates on the radiation dose distribution was more pronounced than that of titanium plates under the same conditions.

Titanium and its alloys are also often used in oral implants, which may have an impact on the radiation dose to patients with nasopharyngeal tumors. Lin et al (24) observed that titanium used as a dental implant in head-and-neck volumetric modulated arc therapy (VMAT) had clinically significant effects on the dose. They used the MC and TPS methods and observed that at a distance of 2 mm from the implant surface, the upstream dose increased by 0.8%; by contrast, the downstream dose was attenuated by 10%. In addition, Ozen et al (28) implanted titanium dental implants of different diameters and lengths into the human mandible and irradiated them with 6 MV X, 25 MV X and Co-60 gamma rays. At the proximity to the titanium, the different sizes of titanium implants increased the dose of Co-60 gamma rays by 17-21%, and the same dose was increased by 17%. However, for 6 MV and 25 MV X-ray irradiation, the dose increased by 17-18% and 15-16%, respectively. Therefore, it could be inferred that the dose increase of 25 MV energy X-rays was slightly lower than that of other energy rays. Nevertheless, there was no significant difference in the dose effect due to the difference in implant size.

Nitinol, a titanium alloy, is widely used to fabricate several types of stents, such as esophageal and tracheal stents. Abu Dayyeh et al (19) used a solid water phantom irradiated with 6, 10 and 18 MV photons. The nitinol stent and wall stent were used; the anterior surface dose was increased by 4.1, 7.1 and 3.2%, respectively; and the other nitinol stent ultraflex was irradiated with the anterior surface dose enhancements of 4.7, 6.1 and 3.7%, respectively. The stainless-steel stent was also used in the aforementioned study, and the effect of the different stent materials on the radiation dose was similar. The main determinant of the dose effect in metallic stents was not the stent material, but the mesh density of the stent. From extensive studies, it was observed that the widely used implants made of nitinol could reduce the effect on RT dose more than the stainless-steel stents; however, caution should be exercised when choosing metal implants because the shape of the stent itself can also affect radiation dose.

In addition to the commonly used stainless steel, titanium and its alloys, several other metals such as gold, ZrO2, and Al₂O₃ are used as materials for artificial implants. These metallic materials are more widely used in the field of dental implants. As reported by Akyol et al (8) the calculated dose increment by MC simulation in front of a dental implant was 15.5 and 3.3% for ZrO2 and Al₂O₃, respectively. The dose decrease behind the dental implant for ZrO2 and Al₂O₃ was 22.2 and 7.0%, respectively. The authors also calculated the change in dose after implantation of metal implants such as titanium. The aforementioned study thus revealed that the density of the implants has an effect on the dose increase in the front of the material, with the higher density increasing the dose to the front surface.

A temporary tissue expander (TTE) is commonly used in patients who require post-mastectomy RT to maintain breast shape and create space between the chest wall and skin. TTE contains an internal metallic port (IMP) used as the injection port for saline injection, which is usually composed of high-density rare-earth magnets, and inevitably perturbs the RT dose (33-35). Using a film dosimetry phantom experiment, Shankar et al (36) observed that the dose attenuation measured at a depth of 22 mm was 22% when irradiated with a single photon beam of 6 MV light. When the energy was raised to 15 MV, the dose attenuation at the same depth was 16%. Furthermore, Gee et al (37) performed in vitro water phantom measurements, measuring the dose distribution at 0.5, 50.0 and 100.0 mm downstream from the IMP, and revealed that the angle of the rays to the IMP had a different effect on the dose, with a 28% metric attenuation when the rays were parallel to one another and a 16% dose attenuation when they were perpendicular to one another. Various degrees of dose reduction were observed downstream in IMP studies (38,39); however, some researchers consider that such dose reduction falls into the saline of TTE and does not significantly affect the surrounding tissues (40).

Various strategies to minimize metal artifacts and improve image quality techniques have been investigated and developed over the years. Dual energy CT is a common method to reduce metal artifacts. It was reported to reduce beam hardening artifacts between 95 and 150 kilo electron volt levels (44,45). Additionally, the use of iterative metal artifact reduction algorithms can reduce metal artifacts and improve dose calculation accuracy, which enables the precise irradiation of tumors (42,46). These techniques are based on projection data and the image-based metal segmentation method that was used as a start (47). There are also commercially available techniques to minimize metal artifacts such as iterative metal artifact reduction (IMAR; Siemens Healthineers) (48), O-MAR (Philips Medical Systems, Inc.) (49), single-energy metal artifact reduction (SEMAR; Toshiba Medical Systems; Canon Medical Systems) (50), and smart metal artifact reduction (Smart MAR (General Electric Healthcare) (51,52). The technique based on projection data and image-based metal segmentation method was used recently, and VM imaging with projection-based material decomposition algorithm can not only reduce metal artifacts effectively, but also simultaneously prevent object blurring at the metal artifact position and image distortion of the metal implants (53,54). Ceccarelli et al (55) considered that combining information from virtual monoenergetic reconstructions and MAR software images could be the best way to solve the issue of metal artifacts on CT images. Those tools were helpful in reducing metal artifacts; however, improved methods need to be further explored. First of all, most of the current research is carried out on the phantom, and it is necessary to verify the effectiveness of its application in the human body, so as to provide a reliable basis for clinical application. In addition, numerous reconstruction algorithms are time-consuming and need to be further optimized to improve the reconstruction efficiency. Finally, the algorithm based on deep learning to reduce metal artifacts has gradually attracted attention. Compared with non-machine learning algorithms, it has advantages in reducing signal-to-noise ratio. In the future, combining big data, deep learning and digital twin technology with increasingly enhanced computer algorithms may improve methods that reduce metal artifacts.

Using the aforementioned techniques, oncologists can obtain a relatively precise processed CT image, making target delineation more accurate. However, it is only through an accurate and fast dose distribution calculation that oncologists can be more confident in RT delivery and avoid unnecessary harm to patients in advance. The algorithm for simulating photon dosage focuses on modeling the deposition pattern of X-rays generated by a linear accelerator in the patient. The common dose calculation algorithms included PBC, analytical anisotropic algorithm (AAA), collapsed cone convolution (CCC) and MC. PBC is the simplest and fastest kernel-based dose computation method. Kernel-based algorithms make use of kernels and ray tracing to model the dose deposition resulting from interactions at a given point. The kernel represents the spread of energy resulting from an interaction at a given point or line, and the ray tracing algorithm represents the energy that passes through the tissue from the energy source. AAA is a convolution-based algorithm which was released in 2005 and implemented in the Eclipse (Varian Medical Systems, Inc.) Integrated TPS (56). CCC algorithm uses one or more-point kernels rather than a line kernel which could accurately model beam hardening as the beam traverses the medium for multiple point kernels. MC is a method of finding numerical solutions to a problem by random simulation which may be used to compute dose distributions by simulating the interactions of a large number of particles (including photons, electrons and protons), as they travel through a medium. It is both the most accurate and computationally intensive method of dose calculations on account of large number of simulated interactions at an atomic level (57).

PBC is only suitable for homogeneous media; its accuracy in non-homogeneous media is poor, and therefore, there are limitations to using it to simulate dosage in the presence of metal implants. A CCC algorithm was used in several TPSs because of its accuracy in homogeneous tissues. In a study by Panettieri et al (58), for the modeled 6-MV photon beams, both the PBC algorithm and the AAA tended to underestimate the absorbed dose in the build-up region compared with the MC results. Paulu and Alaei (59) studied the results of three common dose calculation algorithms in the presence of a hip prosthesis. The aforementioned study found that near the surface of the prosthesis for all energies, a Pinnacle collapse cone convolution algorithm created a 5-22% higher measured dose than calculated, and for the Eclipse Acuros XB and the Eclipse AAA the overestimation of dose was 2-23% and 6-25%, respectively. MC methods are regarded as the ‘gold standard’ for patient dose calculations and are widely used in clinical practice. Ade and Plessis (60) revealed that the MC algorithm used in Monaco was significantly more accurate than the CCC algorithm used in XiO. As Parenica et al (61) reported, the CCC algorithm in Pinnacle demonstrated a significant 9.2% error in calculating the dose and for the MC algorithm in the Monaco TPS the error was 3.6%. Therefore, to the best of our knowledge, the MC algorithm can be used to calculate the dose for the tumor and its surroundings more accurately in the presence of metal implants. Additionally, it can be used as a second check to ensure the accuracy of the RT plan. However, the MC algorithm has some drawbacks, such as the long time period required for computation and the statistical noise when the number of simulated particles is insufficient. Further optimization of the simulation algorithm is needed. For example, in heterogeneous medium, especially at the junction of different density materials, the accuracy of the measurement simulation algorithm needs to be improved. To improve application of these algorithms to clinical practice, it is necessary to further optimize the computational complexity and reduce the computational time.