In this study, we used the improved version of the POWER breathe model, POWER breathe® MEDIC PLUS, to evaluate the combination of IMT with individual training conducted during outpatient rehabilitation. PImax and 6MWD significantly increased. Regarding the Tdi, resting inspiration, resting expiration, and maximal inspiration also increased significantly after training. The Borg Scale scores and RR were significantly lower during the 1-min recovery period following the 6MWT. The loading pressure increased from 20–50% of the PImax, and IMT was performed 30 times, twice a day, for 2 months. IMT load pressure was classified as follows: low-strength load (low load), < 30% of the PImax; moderate-pressure load (medium load), 30% to < 60% of the PImax; and high-strength pressure load (high load), ≥ 60% of the PImax [35–37]. For IMT, the optimum load pressure, at which an effect is reliably obtained, is a medium load of ≥ 30% of the PImax [35, 38]. At present, performing IMT for 15 min twice a day at a load pressure of ≥ 30% of the PImax is a standard practice for patients with COPD . However, it was reported that IMT focused on the number of sessions, and not on the duration, and that the maximum inspiratory pressure increased during one session, in which training was performed 30 times . Considering previous reports [35, 38], in this study we increased the load pressure from 20–50% and asked the participants to perform 30 repetitions twice a day. Interestingly, improvements were also observed in the PImax and 6MWD, similar to the results of the meta-analysis conducted by Beaumont et al. .
An ultrasound imaging system has recently been established as a method for evaluating diaphragm movement, in which Tdi can be measured in the B mode [39–41]. Baria et al.  reported that there was no significant difference in Tdi between patients with COPD and healthy individuals, and also reported a lower limit of normal muscle thickness at rest (i.e., 1.5 mm) in both healthy individuals and patients with COPD.
Furthermore, it has been reported that Tdi increases with an increase in pulmonary volume in healthy individuals [43, 44] and that Tdi decreases significantly in diaphragmatic paralysis . In our study, the resting muscle thickness was 1.39 mm, which was comparable to that reported in a previous study . Additionally, there was a significant increase in resting inspiration, resting expiration, and maximal inspiration after training. This resulted from an increase in FVC and FEV1.0; therefore, in our view, training improved respiratory function and increased Tdi.
In this study, the Tdi was measured using ultrasound before and after IMT, and it was confirmed that IMT affected the respiratory muscle strength, respiratory function, and Tdi. DiNino et al.  reported that Tdi assessment could be a criterion for ventilator weaning. Thus, we consider that Tdi evaluation may be used as an index for respiratory rehabilitation in the future.
In an ATS/ERS systematic review , there was a moderate to strong correlation (coefficient .38–.85) between the 6MWD and physical activity and a strong correlation between low 6MWD (300–450 m) and high mortality; however, prognosis was poor when the 6MWD was < 300 m. The results of our study demonstrated that 6MWD increased significantly after training; although the 6MWD was classified as low, an increase from 354 to 384 m was observed. A further increase in 6MWD is desirable.
The CAT is an eight-item questionnaire that can be used to comprehensively evaluate patients’ HRQOL, and has a clear and strong correlation with St George's Respiratory Questionnaire, which is widely used in the evaluation of quality of life in clinical trials . It has been translated and used in several countries worldwide, including Japan; the same correlation was shown for the Japanese version, which was previously validated . There were no significant differences in the CAT scores before and after training in our study. These results suggested that although the performance of IMT for 2 months improved the respiratory muscle strength and 6MWD, it did not lead to improvement in HRQOL.
The normal range of P0.1 is 1–2 cmH2O [28, 48], and it increases with increased respiratory central activity. In patients with COPD, the value is much higher than that in healthy individuals [49, 50], and increases with COPD severity . The P0.1/PImax index is obtained by normalizing the P0.1 based on differences in inspiratory muscle strength among individuals. P0.1 and P0.1/PImax are used as prediction indicators for weaning  and are regarded as a more accurate predictor compared to PImax . A recent study using P0.1 for COPD showed that P0.1 and P0.1/PImax significantly increased when the severity increased in various relaxation positions ; interestingly, we obtained similar results. In our study, there was no significant difference in P0.1 before and after training, but a significant decrease in P0.1/PImax was observed, reflecting the increase in the PImax. In addition, as no significant difference was observed in P0.1, we considered that the improvement in respiratory function by IMT was probably attributed to improvements in peripheral rather than in central functions. This was also supported by a previous study using P0.1 before and after manual chest wall compression in patients with COPD, which showed that VO2, VCO2, and dyspnea severity significantly decreased after chest compression, whereas P0.1 and P0.1/PImax did not change . Improvement in respiratory function and dyspnea because of manual chest wall compression may be attributed to peripheral rather than central improvement .
Dyspnea improvement is also important in respiratory rehabilitation. The dyspnea sensing mechanism includes not only chemoreceptors and mechanoreceptors, but also motor command  and neuromechanical dissociation [56, 57], involving various factors, such as motor outputs and sensory projections in the central nervous system . Polkey et al.  reported that classic IMT may control muscle and central circuits and is particularly useful for patients with neurological disorders and for those with reduced lung volume because of stroke . However, it remains unclear whether dyspnea improvement and IMT effects are central or peripheral. In this study, during the 1-min recovery period following the 6MWT, the Borg Scale scores for dyspnea and lower limb fatigue and the RR significantly decreased. Furthermore, a negative correlation was found between the 6MWD and the Borg Scale dyspnea score (after the 6MWT), suggesting that dyspnea is one of the factors that limit performance in the 6MWT. Therefore, dyspnea improvement is necessary for continuous movement. Moreover, the fact that the PImax, 6MWD, and Tdi significantly increased while P0.1 did not change after IMT further suggested that the IMT effects are attributed to improvements in peripheral factors rather than to improvements of central factors.
One of the strengths of this study was the finding that the improvement in respiratory function by IMT was likely to be attributed to peripheral factors. Elucidating whether improvements in dyspnea and respiratory function are attributed to central or peripheral components would contribute to the design of more effective respiratory rehabilitation intervention methods, help determine more accurately the effects of respiratory rehabilitation, and help identify more appropriate candidates for rehabilitation.
This study had several limitations, including the small sample size, outpatient-based IMT, lack of a control group, patient compliance validation, and technical outcome measurements. In a future research, we will further increase the number of participants and use a control group to examine the differences among different IMT devices.